Re: [s3ql] Upgrade from S3QL 1.15 to 2.7
On 02/27/2014 09:04 AM, Dan Johansson wrote: One question though, can I set the --no-ssl globally in some config-file? That way I do not have to update my scripts that implement s3ql. No, that's not possible at this time. Sorry. Best, -Nikolaus -- Encrypted emails preferred. PGP fingerprint: 5B93 61F8 4EA2 E279 ABF6 02CF A9AD B7F8 AE4E 425C »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/groups/opt_out.
Re: [s3ql] Mount.s3ql hangs after 403 responses
Nicola Cocchiaro nicola.cocchi...@gmail.com writes: Nikolaus, A while ago we had discussed an llfuse exception in a thread called strange llfuse exception that was not being correctly formed due to a Cython bug. With the workaround in llfuse 0.40 the issue with the exception was resolved, but the same triggering event of Google Storage returning 403 intermittently still triggers an exception that results in bad behavior. Specifically, after an AccessDenied exception (or two as in the logs below), mount.s3ql hangs and with it all file system operations also hang (e.g., accessing the file system extended attributes or running the 'ls' and 'df' commands). [...] This triggers another AccessDenied exception, but that's where the log ends. At this point S3ql is no longer responsive, 'ls' and 'df' hang, and any other file system operation also seems to hang. This also happened consistently with all S3ql processes I had running on other boxes that happened to be uploading data in the time window when Google intermittently returned 403. Thanks for the report. I'll try to reason out what's happening here. Could you please report this issue at https://bitbucket.org/nikratio/s3ql/issues to make sure it doesn't get lost? Also, if you can reproduce this issue (or encounter it again) and mount.s3ql hangs, could you please try to obtain a Python stacktrace as explained on https://bitbucket.org/nikratio/s3ql/wiki/Providing%20Debugging%20Info? That'd make it much easier to diagnoses what's going on. Question is, is this due to the timing of those (repeated) exceptions and how S3ql handles them, or a FUSE bug? No, this is probably an S3QL bug. Best, Nikolaus -- Encrypted emails preferred. PGP fingerprint: 5B93 61F8 4EA2 E279 ABF6 02CF A9AD B7F8 AE4E 425C »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/groups/opt_out.
Re: [s3ql] Possible to run S3QL on Debian Wheezy (Python 3.2)?
Tor Krill tor.kr...@gmail.com writes: Hi all, I really would like to run S3Ql 2.x on Debian Wheezy. Is this possible? Yes, you just need to install Python 3.3. Looking at the PPA for Ubuntu S3QL requires Python 3.3 but Wheezy only has Python 3.2. I investigated the possibilities to backport Python 3.3 to Wheezy but that seems like quite an undertaking. Hmm. I think the last time I checked, it was enough to just recompile the jessy packages for wheezy: apt-get source python3.3/testing apt-get build-dep python3.3 (cd python3*; dpkg-buildpackage -us -uc) Have you tried that? So to summarize, is it possible to run S3QL 2.x with Python 3.2? If not, what is the problem? S3QL 2.x won't run on Python 3.2 without changes. There are some missing modules (faulthandler, lzma), some exceptions don't have names yet (http://docs.python.org/3/whatsnew/3.3.html#pep-3151), yield from does not exist, contextlib.ExitStack is missing, hmac.compare_digest is missing, and probably some more things. Best, -Nikolaus -- Encrypted emails preferred. PGP fingerprint: 5B93 61F8 4EA2 E279 ABF6 02CF A9AD B7F8 AE4E 425C »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: Private message regarding: [s3ql] HTTP timeouts
Hi Nicola, Nicola Cocchiaro nicola.cocchi...@gmail.com writes: The reason I originally asked was due to seeing some outbound connections not completing but just hanging, until using umount.s3ql would let them return with a TimeoutError (no less than 15 minutes later in all cases seen). I was not able to dig much deeper at the time, but to experiment more I put together a simple patch to add a configurable socket timeout to all S3ql tools that may make use of it. I've attached it if you'd like to consider it. Thanks for the patch! I'm generally rather reluctant to add new command-line options unless they are absolutely crucial. The problem is that the number of possible configurations (and potential interaction bug) goes up exponentially with every option. In this case, I am not sure I fully understand in which situation this option is intended to be used (so I'm pretty sure that a regular user wouldn't know when to use it either, which is always a bad sign for a command line argument). Could you give some additional details on that? For example, if I'm not happy with the system timeout (which seems to be 15 minutes in your case), shouldn't this be adjusted on the os level as well? And if not, is there really a need to make the timeout configurable rather than having S3QL simply use an internal default? Best, -Nikolaus -- Encrypted emails preferred. PGP fingerprint: 5B93 61F8 4EA2 E279 ABF6 02CF A9AD B7F8 AE4E 425C »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: Private message regarding: [s3ql] HTTP timeouts
Nicola Cocchiaro nicola.cocchi...@gmail.com writes: On Tuesday, March 11, 2014 8:06:42 PM UTC-7, Nikolaus Rath wrote: Hi Nicola, Nicola Cocchiaro writes: The reason I originally asked was due to seeing some outbound connections not completing but just hanging, until using umount.s3ql would let them return with a TimeoutError (no less than 15 minutes later in all cases seen). I was not able to dig much deeper at the time, but to experiment more I put together a simple patch to add a configurable socket timeout to all S3ql tools that may make use of it. I've attached it if you'd like to consider it. Thanks for the patch! I'm generally rather reluctant to add new command-line options unless they are absolutely crucial. The problem is that the number of possible configurations (and potential interaction bug) goes up exponentially with every option. In this case, I am not sure I fully understand in which situation this option is intended to be used (so I'm pretty sure that a regular user wouldn't know when to use it either, which is always a bad sign for a command line argument). Could you give some additional details on that? For example, if I'm not happy with the system timeout (which seems to be 15 minutes in your case), shouldn't this be adjusted on the os level as well? And if not, is there really a need to make the timeout configurable rather than having S3QL simply use an internal default? The problem is that there was no apparent system timeout, or those connections did not seem to be timing out on their own in the cases observed; I did not have the means at the time to figure out why exactly this would happen, but the root cause of all this was a temporary malfunction on the Google Storage side. The 15 minutes come from a different process which had its own timeout (15 minutes in fact) for allowing S3ql to unmount; in response to the timeout firing it called umount.s3ql again, and that in turn seemed to allow the connections to be recognized as timed out (possibly in response to sending SIGTERM to the mount.s3ql process? This was my theory after looking at the timing from a number of logs but again, unfortunately I do not have 100% solid evidence that this was the reason). Hmm. I don't think this is likely. umount.s3ql does not send SIGTERM. It sets a termination flag that is checked for the in main file system loop. So just calling umount.s3ql would not result in any pending socket operations to terminate. Is there any way to reproduce the problem you had? A static, internal default may be enough and in fact it helped when I first tried it, but more advanced users may want to adapt the timeout to their own use case when relying on the OS doesn't seem to help like in the cases observed. I understand the reluctance to add more options and increase complexity for all users, but I thought I'd share this patch for consideration, perhaps as an extra option in a set of clearly marked advanced options. Understood, thanks. I don't want to rule out such an option yet, but if we add it, add the very least there should be some documentation explaining what exactly the option does, and when it should be used. At the moment, this seems rather unclear to me (see above). Best, -Nikolaus -- Encrypted emails preferred. PGP fingerprint: 5B93 61F8 4EA2 E279 ABF6 02CF A9AD B7F8 AE4E 425C »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] encryption + compression + deduplication?
On 03/14/2014 08:11 AM, aweber1nj wrote: Do all the features in the subject work well together? That is, can I enable all three options and assume: 1. All my data will be stored encrypted. 2. Data is compressed before storing. 3. Identical blocks (of original content, pre-compression and encryption I would assume) are only stored once (correctly de-duplicated)? Yes, they all work together. The order is: (1) deduplication, (2) compression, (3) encryption. Best, -Nikolaus -- Encrypted emails preferred. PGP fingerprint: 5B93 61F8 4EA2 E279 ABF6 02CF A9AD B7F8 AE4E 425C »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] limiting bandwidth?
Hi, On 03/26/2014 06:23 AM, Balázs wrote: Hi Niko, I was wondering if there is any way to limit / control the available bandwidth that is available to the s3ql backend for its upload to the cloud. The amount of data I have to move on a daily basis hangs into the production hours, and other services suffer. I was hoping I can somehow limit the upload total bandwidth, so that the syncs can run during the day and finish... Not in S3QL itself, but the Linux IP stack is your friend :-). As an alternative workaround, maybe it helps to use fewer compression threads and lzma compression at a high level? That ought to slow things down quite a bit... Best, -Nikolaus -- Encrypted emails preferred. PGP fingerprint: 5B93 61F8 4EA2 E279 ABF6 02CF A9AD B7F8 AE4E 425C »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] s3ql quotas
Hi PA, When replying to emails on this list, please do not put your reply above the quoted text, and do not quote the entire message you're answering to. This makes it unnecessarily hard for other readers to understand the context of your email. Instead, please cut quoted parts that are not relevant to your reply, and insert your responses right after the points you're replying to (as I have done below). Thanks! PA Nilsson p...@zid.nu writes: A follow up question on this. When mounting a file system using an s3c backend, running 'df' will report that the filsystem has 1TB size. There is no such information coming from the backend, but is is easily made available. Can information on this somehow be propagated? As far as I know, there is no such information from most backends. Google Storage, Amazon S3, OpenStack et al all have effectively unlimited storage. I believe only the local backend could effectively report a capacity. Which backend do you have in mind? But even if we got a number from the backend, it's not clear what we should report to df. How do we take into account compression and deduplication? Best, -Nikolaus -- Encrypted emails preferred. PGP fingerprint: 5B93 61F8 4EA2 E279 ABF6 02CF A9AD B7F8 AE4E 425C »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] CentOS 6.5 installation with failures - NameError: name 'xrange' is not defined
On 04/01/2014 07:45 AM, Randy Black wrote: [...] -- Download s3ql and install cd /tmp; curl -O http://s3ql.googlecode.com/files/s3ql-1.14.tar.bz2 tar -jxvf s3ql-1.14.tar.bz2; cd s3ql-1.14 python3.3 ./setup.py install If you have Python 3.3, you don't need to use the (outdated) S3QL 1.x branch. Use the up-to-date 2.x branch instead (current version is 2.8), and everything should be fine. Best, -Nikolaus -- Encrypted emails preferred. PGP fingerprint: 5B93 61F8 4EA2 E279 ABF6 02CF A9AD B7F8 AE4E 425C »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Test errors s3ql-2.8.1 on Centos 6.5 64bit
Cristian Manoni nethc...@gmail.com writes: and now the test: # python3 runtests.py tests after some succesful tests: tests/t2_block_cache.py:114: cache_tests.test_destroy_deadlock FAILED FAILURES ___ cache_tests.test_destroy_deadlock Traceback (most recent call last): File /root/s3ql/s3ql-2.8.1/tests/t2_block_cache.py, line 163, in test_destroy_deadlock self.cache.destroy() File _pytest.python, line 1009, in __exit__ pytest.fail(DID NOT RAISE) File _pytest.runner, line 456, in fail raise Failed(msg=msg, pytrace=pytrace) Failed: DID NOT RAISE ! Interrupted: stopping after 1 failures ! 1 failed, 104 passed, 97 skipped in 5.15 seconds What is missing or what is wrong? Can you help me? Hmm. Does it by any chance help if you apply the following patch? diff --git a/tests/t2_block_cache.py b/tests/t2_block_cache.py --- a/tests/t2_block_cache.py +++ b/tests/t2_block_cache.py @@ -156,6 +156,7 @@ # Shutdown threads llfuse.lock.release() +time.sleep(10) try: with catch_logmsg('Unable to flush cache, no upload threads left alive', level=logging.ERROR): Thanks, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] dugong.UnsupportedResponse: No content-length and no chunked encoding
On 04/14/2014 10:06 AM, Adam Watkins wrote: This comes from another server running Debian 7.4 and using the S3QL 1.11 package: This server _isn't_ running on my home network, and I've no reason to believe that it's connection is particularly unreliable. [] Thanks! Could you please file this as a bug on https://bitbucket.org/nikratio/s3ql/issues? I'll see what I can do. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
[s3ql] [ANNOUNCE] S3QL 1.18 has been released
Dear all, I am pleased to announce a new maintenance release of S3QL, version 1.18. Please note that this is only a maintenance release. Development of S3QL takes place in the 2.x series. The 1.x releases receive only selected bugfixes and are only maintained for older systems that do not support Python 3.3. For systems with Python 3.3 support, using the most recent S3QL 2.x version is strongly recommended. From the changelog: 2014-04-26, S3QL 1.18 * Fixed a problem with mount.s3ql crashing with `KeyError` under some circumstances. * Fixed a problem with mount.s3ql (incorrectly) reporting corrupted data for compressed blocks of some specific sizes. Many thanks to Balázs for extensive debugging of this problem. * s3qlrm now also works when running as root on a file system mounted by a regular user. Please report any bugs on the mailing list (s3ql@googlegroups.com) or the issue tracker (https://bitbucket.org/nikratio/s3ql/issues). Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« signature.asc Description: OpenPGP digital signature
[s3ql] [ANNOUNCE] S3QL 1.18.1 has been released
Dear all, I am pleased to announce a new maintenance release of S3QL, version 1.18.1. Please note that this is only a maintenance release. Development of S3QL takes place in the 2.x series. The 1.x releases receive only selected bugfixes and are only maintained for older systems that do not support Python 3.3. For systems with Python 3.3 support, using the most recent S3QL 2.x version is strongly recommended. From the changelog: 2014-04-28, S3QL 1.18.1 * No changes in S3QL itself. * The S3QL 1.18 tarball accidentally included a copy of the Python dugong module. This has been fixed in the 1.18.1 release. 2014-04-26, S3QL 1.18 * Fixed a problem with mount.s3ql crashing with `KeyError` under some circumstances. * Fixed a problem with mount.s3ql (incorrectly) reporting corrupted data for compressed blocks of some specific sizes. Many thanks to Balázs for extensive debugging of this problem. * s3qlrm now also works when running as root on a file system mounted by a regular user. Please report any bugs on the mailing list (s3ql@googlegroups.com) or the issue tracker (https://bitbucket.org/nikratio/s3ql/issues). Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« signature.asc Description: OpenPGP digital signature
Re: [s3ql] Detect that a bucket is mounted elsewhere?
On 05/09/2014 05:55 AM, andycr...@gmail.com wrote: Folks, I am using s3ql-1.17 (on CentOS5 6) and I have a question. From each system, it does fsck.s3ql to see if the bucket was previously formatted, and if not it does mkfs.s3ql. This sounds dangerous. Can't you create the file system once ahead of time? If not, why (and how) do you use fsck.s3ql for this purpose? Wouldn't it be enough to always call mkfs.s3ql without --force? The issue is that if the bucket happened to be mounted on another system, the second one blocks forever in fsck.s3ql. What do you mean by that? Do you mean it's waiting for input? In that case, just redirect from /dev/null. Is there a way to detect that the bucket is mounted elsewhere, so I can avoid this? There is no standalone program, but you should be able to just try to mount or fsck it. If the file system is mounted elsewhere, this should fail with an error message. Whether this is 100% reliable depends on your backend, e.g. with Amazon S3 you are not guaranteed to always get current data, so you can never be absolutely sure that the file system is not mounted elsewhere. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Force fsck to continue?
PA Nilsson p...@zid.nu writes: It's a directory, not a file, and it is created when mount.s3ql starts. If this directory (with its contents) disappears if you reboot the system several minutes later, you have a real problem. If I reboot several minutes later, the directory is there and everything works. I need to force the reboot within seconds after the mount process, or maybe even during it while the metadata is read/written from the server. Ah, you didn't say that before. In that case there might be an easy fix that can be implemented in S3QL. Stay tuned, I'll send you a patch soonish. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Force fsck to continue?
PA Nilsson p...@zid.nu writes: So my thinking is that this is a problem that we have with our flash based file system. The file is simply not yet written to flash. This will be running on an non maintained system with no possibility for user interaction. If there is a high chance that the system will be power-cycled without a proper shutdown, you may want to apply the following patch as swell (in addition to the patch from my other mail). It reduces the likelyhood of metadata data corruption on power loss, but also reduces performance (so this is not going to go into the official S3QL code): diff --git a/src/s3ql/database.py b/src/s3ql/database.py --- a/src/s3ql/database.py +++ b/src/s3ql/database.py @@ -32,7 +32,7 @@ # locking_mode to EXCLUSIVE, otherwise we can't switch the locking # mode without first disabling WAL. 'PRAGMA synchronous = OFF', - 'PRAGMA journal_mode = OFF', + 'PRAGMA journal_mode = NORMAL', #'PRAGMA synchronous = NORMAL', #'PRAGMA journal_mode = WAL', Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] fsck.s3ql Deleted spurious object
On 06/21/2014 05:38 PM, Warren Daly wrote: Could you please zip all the mount.log* and fsck.log* files and put them somewhere on the web? I'd like to see when this problem started appearing. http://www.invisibleagent.com/20mntfsk.tar $ wget http://www.invisibleagent.com/20mntfsk.tar --2014-06-21 17:44:26-- http://www.invisibleagent.com/20mntfsk.tar Resolving www.invisibleagent.com (www.invisibleagent.com)... 108.162.196.30, 108.162.197.30, 2400:cb00:2048:1::6ca2:c41e, ... Connecting to www.invisibleagent.com (www.invisibleagent.com)|108.162.196.30|:80... connected. HTTP request sent, awaiting response... 403 Forbidden 2014-06-21 17:44:27 ERROR 403: Forbidden. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] memory usage and --compress options
On 06/30/2014 10:22 PM, Brice Burgess wrote: I went ahead and used the --compress none flag to disable compression. I had to re-create the fileksystem to get it working (as it was previously mounted with LZMA compression, and thus rightly retains this setting upon remounting it). It should not. What makes you think it retained the setting? Mounting the (existing) file system with --compress alg should change the compression of all new objects. There is no need to worry about decompressing the old LZMA compressed objects, that only takes a small amount of memory. Can anyone forsee issues running with --compress none? Obviously traffic is increased; but memory (and CPU) is tight on these small VMs. Would bzip/zlib be a better option? Yes. Both require orders of magnitude less memory than lzma, so I'd definitely try one of those instead of not compressing at all. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] ports used by s3ql for firewall?
On 07/11/2014 09:36 AM, Andy Cress wrote: Which ports do I need to make sure are open in a firewall in order to use s3ql mount points? S3QL uses only outgoing connections. By default, it uses only TCP ports 80 and 443 (for HTTP and HTTP via TLS). If you specify a different port number in your https_proxy environment variable, or in the storage url for your backend, S3QL will obviously use that port as well. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] High CPU usage when writing large files
Jimmy Tanzil jimmy.tan...@ichthysmedia.com writes: I am having high CPU load issues when writing large files (above 5 GB) into the mounted s3ql filesystem. Could you quantify that? How do you measure it, and how does it depend on file size? Is there really a jump at 5 GB, or is it rather a smooth increase? My mount command: mount.s3ql --compress none --debug all --allow-other s3://bucket /backup15 Using --debug all is certainly a way to eat up a lot of CPU time. But I assume the problem also happens with debugging disabled? Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] High CPU usage when writing large files
Hi Jimmy, When replying to emails on this list, please do not put your reply above the quoted text, and do not quote the entire message you're answering to. This makes it unnecessarily hard for other readers to understand the context of your email. Instead, please cut quoted parts that are not relevant to your reply, and insert your responses right after the points you're replying to (as I have done below). Thanks! Jimmy Tanzil jimmy.tan...@ichthysmedia.com writes: I am having high CPU load issues when writing large files (above 5 GB) into the mounted s3ql filesystem. Could you quantify that? How do you measure it, and how does it depend on file size? Is there really a jump at 5 GB, or is it rather a smooth increase? When writing less than 100 MB data, it takes only a few seconds and it finished very quick. But when the file size is larger, the CPU usage keeps climbing and stays there until the file write is complete. I created a 2 minute video to show you what I mean: http://youtu.be/-HMac-Hr3bw Sorry, that's not helpful at all. Please provide quantitative data in text form. The first file copied in the video is 80 MB in size The second file copied in the video is 5 GB in size How long does each copy take? Do both files with into the local cache? Are taking into account the time it takes to flush the cache? Is this normal? Is what normal? Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Time Skewed error repeats forever
On 07/16/2014 01:33 PM, Andy Cress wrote: I encountered a client s3ql system with its system date/time set at about 4 hours before the actual time. On that system, attempting to do fsck.s3ql to the cloud resulted in an infinite set of retries with these messages: Encountered RequestTimeTooSkewedError exception (RequestTimeTooSkewed: The difference between the request time and the current time is too large.), retrying call to Backend.open_read Is there a way to limit the retries on this? If it returned, it would be easy to check for this condition. Or is there a different way to check for time skew before calling fsck.s3ql? No, but I think S3QL should really not retry on this error at all. Could you report a bug? https://bitbucket.org/nikratio/s3ql/issues/new Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] High CPU usage when writing large files
On 07/16/2014 01:27 PM, Jimmy Tanzil wrote: Is what normal? The CPU being utilized more than 100% during writing larger files. My concern is not the timing, the timing is excellent. I am more concerned about the CPU showing above 100% when writing one 5 GB file. Is there any other tricks that can be done to lower that? I do not understand your concern. Why would you want to artifically slow down S3QL by forcing it not to use all the available CPU cycles? You can add a lot of sleep(1) statements to the write() operation in fs.py. That should bring CPU utilization down to 1 %. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Enabling AWS server-side encryption in S3QL?
Mark Mielke mark.mie...@gmail.com writes: On Sat, Jul 19, 2014 at 11:19 PM, Nikolaus Rath nikol...@rath.org wrote: It has been requested (https://bitbucket.org/nikratio/s3ql/issue/62/add-support-for-aws-server-side-encryption) that I enable AWS server-side encryption in S3QL. I am ambigious on the matter. On one side, there does not seem to be any technical drawback. On the other side, there does not seem to be any (significant) technical advantage either, so I'm still hesitant to enable this without a good reason. If anyone has some thoughts on the question, please chime in. Presuming I understand how it works... I think having AWS perform the encryption partially defeats the purpose of encryption. [..] Oh, absolutely. S3QL would continue to do client-side encryption as before. The question is: is there any good reason to activate (or not activate) server-side encryption *in addition* to that. I am very irritated by this feature for the same reasons you mentioned above. It doesn't seem to add any value, but it doesn't seem to have any drawbacks either. So why is Amazon even requiring a choice at all? Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] High CPU usage when writing large files
Jimmy Tanzil jimmy.tan...@ichthysmedia.com writes: There is no indication of any crash (nor any other problem) in the log file at all. Where did you get the File system appears to have crashed message from? I see it when I try to do umount.s3ql /mountpoint Then I read on your documentation to just do umount /mountpoint which works I then did fsck.s3ql which completes under 1 minute, then I can mount it back successfully. I tried to unmount it in the first place because I cannot access the mounted file system. If I do ls or any command including s3qlstats it just sits there without any response. Is the mount.s3ql process still running? (check with ps) If so, can you obtain a stack trace (https://bitbucket.org/nikratio/s3ql/wiki/Providing%20Debugging%20Info)? Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] High CPU usage when writing large files
On 07/20/2014 04:06 PM, Jimmy Tanzil wrote: Is there an option on the mount command to utilize 80% of the available CPU cores? No, and I don't think that introducing it would make sense. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Using folders in s3ql
On 07/24/2014 02:24 PM, Andy Cress wrote: After checking this out on both Amazon and Google, I can see that the implementation is consistent, but not what I expected. A folder is created by the administrator (e.g. 'myfolder'), and the s3ql data for mybucket/myfolder goes into blocks in the bucket root which have the following naming convention: myfolders3ql_data_* (using s3ql-1.17) I'm curious. Why do those files go into the bucket root instead of into myfolder? There is no such as things as folders in Amazon S3 or Google Storage. The string folder/file is the name of an object, just like folder+file or folderfile. The / character has no special meaning at all (see also http://docs.aws.amazon.com/AmazonS3/latest/dev/UsingObjects.html). You can easily check that as follows: 1. Create an object folder/file1 3. Try to rename folder to folder_new - you will get an error, because there is no object named folder. For convenience, the gsutil command and the AWS Management console allow you to pretend that '/' has any meaning, but it really has not. If you want s3ql to store its objects as myfolders/s3ql_data_*, you need to use myfolder/ as a prefix, not myfolder. Best, Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
[s3ql] [ANNOUNCE] S3QL 2.10 has been released
Dear all, I am pleased to announce a new release of S3QL, version 2.10. From the changelog: 2014-07-27, S3QL 2.10 * The internal file system revision has changed. File systems created with S3QL 2.10 or newer are not compatible with prior S3QL versions. To update an existing file system to the newest revision, use the 's3qladm upgrade' command. It is strongly recommended to run the (new) s3ql_verify command with the --data option at shortly after the upgrade. This is necessary to ensure that the upgrade to the next (2.11) S3QL release will run smoothly. * The User's Guide now contains a description of the possible failure modes of mount.s3ql. * The --debug command line parameter now generates a bit less output by default, and there is an additional --debug-module parameter to activate additional messages. * When using encryption, S3QL now checks that a storage object's key corresponds to the data stored under it. The lack of this check in previous versions allowed an attacker with control over the storage server to interchange blocks within the same file system (which would have resulted in apparent data corruption in the file system). Targeted modification of specific files, however, would have been very unlikely, because the interchange of blocks had to be done blindly (with the attacker not knowing which file any block belongs to, nor what its contents are). Fixes https://bitbucket.org/nikratio/s3ql/issue/52/. * S3QL now aborts immediately instead of retrying if the server storage server reports that the local clock is skewed. * There is a new 's3ql_verify' command. This program retrieves and checks every storage object to ensure that all objects are available and have not been tampered with. In contrast to fsck.s3ql, s3ql_verify does not trust the object list provided by the storage server and actually attempts to download the objects one by one. * S3QL now requires version 3.2 or newer of the dugong module. As usual, the release is available for download from https://bitbucket.org/nikratio/s3ql/downloads Please report any bugs on the mailing list (s3ql@googlegroups.com) or the issue tracker (https://bitbucket.org/nikratio/s3ql/issues). Starting with version 2.0, S3QL requires Python 3.3 or newer. For older systems, the S3QL 1.x branch (which only requires Python 2.7) will continue to be supported for the time being. However, development concentrates on S3QL 2.x while the 1.x branch only receives selected bugfixes. When possible, upgrading to S3QL 2.x is therefore strongly recommended. Enjoy, -Nikolaus -- Encrypted emails preferred. PGP fingerprint: 5B93 61F8 4EA2 E279 ABF6 02CF A9AD B7F8 AE4E 425C »Time flies like an arrow, fruit flies like a Banana.« signature.asc Description: OpenPGP digital signature
Re: [s3ql] possible to upgrade filesystem from v1.12 to 2.9
Nikolaus Rath nikol...@rath.org writes: This bucket was created using s3ql v1.12 so I was a little confused by the Fyle system already at most-recent revision' output. Could this have anything to do with my failed attempt to upgrade it from 1.2 to 2.9 already. (I assume you mean 1.12 in the last line). Outside of that question, I then attempted to upgrade again on my 2.9 host and got the following output: s3qladm upgrade s3://s3ql-test2 Enter backend login: Enter backend passphrase: Enter file system encryption passphrase: Uncaught top-level exception: [..] File /usr/lib/s3ql/s3ql/backends/common.py, line 612, in open_read metadata = self._unwrap_meta(fh.metadata) File /usr/lib/s3ql/s3ql/backends/common.py, line 571, in _unwrap_meta raise ObjectNotEncrypted() s3ql.backends.common.ObjectNotEncrypted That is odd. I tried to reproduce this on my system, but it seems that S3QL 1.12 isn't compatible with the more recent rest of my system anymore. Ok, I set up an old wheezy chroot, and when I try it than upgrading from 1.12 works fine: (wheezy)root@vostro:~# mkfs.s3ql --version S3QL 1.12 (wheezy)root@vostro:~# mkfs.s3ql s3://nikratio-test/upgrade/ Enter backend login: Enter backend password: Before using S3QL, make sure to read the user's guide, especially the 'Important Rules to Avoid Loosing Data' section. Enter encryption password: Confirm encryption password: Generating random encryption key... Creating metadata tables... Dumping metadata... ..objects.. ..blocks.. ..inodes.. ..inode_blocks.. ..symlink_targets.. ..names.. ..contents.. ..ext_attributes.. Compressing and uploading metadata... Wrote 0.00 MiB of compressed metadata. (wheezy)root@vostro:~# ..and on the other system: $ bin/s3qladm --version S3QL 2.9 $ bin/s3qladm upgrade s3://nikratio-test/upgrade/ Enter file system encryption passphrase: Getting file system parameters.. I am about to update the file system to the newest revision. You will not be able to access the file system with any older version of S3QL after this operation. You should make very sure that this command is not interrupted and that no one else tries to mount, fsck or upgrade the file system at the same time. Please enter yes to continue. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Installation fails at self tests
Alexandre Goncalves alexandre@gmail.com writes: Domingo, 3 de Agosto de 2014 4:29:35 UTC+1, Nikolaus Rath escreveu: On 08/02/2014 08:13 PM, Alexandre Goncalves wrote: Hello, I want to install S3QL in my CentOS 6.5 (64bit) box. For that I did: install Python3.4.1 (from source) install setuptools (using pip3.4) install PyCrypto (using pip3.4) install defusedxml (using pip3.4) install sqlite 3.8.5 from source install apsw (using setup.py) [...] /root/temp/s3ql-2.10.1/src/s3ql/deltadump.cpython-34m.so: undefined symbol: sqlite3_compileoption_get Are you sure that you compiled both apsw and s3ql against the sqlite 3.8.5 library that you installed, rather than some other version that may already be present on your system? Try to run 'ldd /root/temp/s3ql-2.10.1/src/s3ql/deltadump.cpython-34m.so' and 'ldd /wherever/it/is/apsw.cpython-34m.so'. The output of your commands: ldd /usr/local/lib/python3.4/site-packages/apsw.cpython-34m.so linux-vdso.so.1 = (0x7fffe33ff000) libpython3.4m.so.1.0 = /usr/local/lib/libpython3.4m.so.1.0 (0x7f9e8bd6c000) libpthread.so.0 = /lib64/libpthread.so.0 (0x7f9e8bb42000) libc.so.6 = /lib64/libc.so.6 (0x7f9e8b7ae000) libdl.so.2 = /lib64/libdl.so.2 (0x7f9e8b5aa000) libutil.so.1 = /lib64/libutil.so.1 (0x7f9e8b3a6000) libm.so.6 = /lib64/libm.so.6 (0x7f9e8b122000) /lib64/ld-linux-x86-64.so.2 (0x003aac40) No mention of sqlite, so you have probably included sqlite directly in apsw. Not elegant, but it works and ensures that you get the version you want. I compiled apsw using sqlite 3.8.5, since I used: python3.4 setup.py fetch --all --sqlite --version=3.8.5 --missing-checksum-ok build --enable-all-extensions install test Yeah, 'fetch' means no dynamic linking. BUT i suspect that I should have used the local compilation!... Correct? Would have been nicer, but it should still work. ldd /root/temp/s3ql-2.10.1/src/s3ql/deltadump.cpython-34m.so linux-vdso.so.1 = (0x7fff88dff000) libpython3.4m.so.1.0 = /usr/local/lib/libpython3.4m.so.1.0 (0x7f3a798ac000) libsqlite3.so.0 = /usr/lib64/libsqlite3.so.0 (0x7f3a7961) libpthread.so.0 = /lib64/libpthread.so.0 (0x7f3a793f3000) libc.so.6 = /lib64/libc.so.6 (0x7f3a7905f000) libdl.so.2 = /lib64/libdl.so.2 (0x7f3a78e5a000) libutil.so.1 = /lib64/libutil.so.1 (0x7f3a78c57000) libm.so.6 = /lib64/libm.so.6 (0x7f3a789d3000) /lib64/ld-linux-x86-64.so.2 (0x003aac40) On s3q, I don't know how to tell the locationof the SQlite 3.8.5, and I think it is picking the other version! Yes, that's probably the problem. You could try something like CFLAGS=-I/sqlite_directory/include -L/sqlite_directory/lib -Wl,-rpath=/sqlite_directory/lib setup.py build_ext Best, Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] S3QL and automount
On 08/03/2014 08:07 PM, Alexandre Goncalves wrote: Hello, I want mount my bucket as needed, so I tried to configure autofs to mount the gs bucket. Tried this, without luck: auto.master: /gs-backup /etc/auto.gs--timeout=300 auto.gs: * -fstype=s3ql,rw,allow_other gs://ideiao-bkp --authfile /root/scripts/.s3ql_authinfo2 Note: Mounting manually, works perfectly: mount.s3ql gs://ideiao-bkp /mnt/1 --authfile /root/scripts/.s3ql_authinfo2 Using 10 upload threads. Autodetected 4034 file descriptors available for cache entries Using cached metadata. Setting cache size to 6935 MB Mounting filesystem... Any suggestions? If you elaborate on what without luck means (maybe even include a specific error message), someone might be able to help you. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Crash s3ql-2.9.1 in Fedora 20 packages with Swift
On 08/06/2014 04:22 PM, motomura wrote: I use s3ql-2.9.1 in Fedora 20 packages with OpenStack Swift (/v1) as backed storage. s3ql crash often. Below the log when s3ql is crashed. Is it a bug of s3ql? - 2014-08-06 17:19:09.067 [pid=674, thread='Thread-6', module='root', fn='excepthook', line=163]: Uncaught top-level exception: Traceback (most recent call last): [...] return method(*a, **kw) File /usr/lib64/python3.3/site-packages/s3ql/backends/s3c.py, line 758, in close headers=self.headers, body=self.fh) File /usr/lib64/python3.3/site-packages/s3ql/backends/swift.py, line 214, in _do_request shutil.copyfileobj(body, self.conn, BUFSIZE) File /usr/lib64/python3.3/shutil.py, line 71, in copyfileobj fdst.write(buf) File /usr/lib/python3.3/site-packages/dugong/__init__.py, line 592, in write eval_coroutine(self.co_write(buf)) File /usr/lib/python3.3/site-packages/dugong/__init__.py, line 1309, in eval_coroutine assert next(crt).poll() File /usr/lib/python3.3/site-packages/dugong/__init__.py, line 604, in co_write raise StateError('No active request with pending body data') Yes, that looks like a bug. Could you try to reproduce that with S3QL 2.10.1 and report it at https://bitbucket.org/nikratio/s3ql/? Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Upgrading from revision 20 to 21...
Serge Victor sergi...@pawlowicz.name writes: The s3ql worked fine. After upgrading from s3ql_2.9+dfsg-2~26~ubuntu14.04.1_amd64.deb to s3ql_2.10.1+dfsg-1~27~ubuntu14.04.1_amd64.deb, I am trying to upgrade revision, as required, which fails: [...] Cycling metadata backups... Backing up old metadata... Encountered ConnectionClosed exception (connection closed unexpectedly), retrying call to Backend.copy for the 3-th time... [...] Are you able to create, mount, modify and umount a new filesystem using 2.10.1? I'd like to narrow down if the problem is specific to the upgrade procedure, or applies to regular usage as well. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Crash s3ql-2.9.1 in Fedora 20 packages with Swift
Hi motomura, On 08/08/2014 01:03 AM, motomura wrote: s3ql crashed in version 2.10.1. Below the log when s3ql is crashed. - 2014-08-08 16:55:01.509 [pid=8686, thread='Thread-6', module='root', fn='excepthook', line=163]: Uncaught top-level exception: Traceback (most recent call last): File [...] Thanks for reproducing. Could you report this at https://bitbucket.org/nikratio/s3ql/issues so it doesn't get lost? Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Unexpected server reply to copy operation (upgrading from 20 to 21)
On 08/14/2014 03:17 PM, Adam Watkins wrote: Upgrading from revision 20 to 21... Unexpected server reply to copy operation: 200 OK Date: Thu, 14 Aug 2014 21:04:15 + Server: RestServer/1.0 Content-Length: 189 Content-Type: application/xml ETag: d7ed6a310a47b69a91e13e655245fac9 Cache-Control: no-cache Connection: close ?xml version=1.0 encoding=UTF-8? CopyObjectResultLastModified2014-08-14T21:04:15.000Z/LastModified ETagquot;d7ed6a310a47b69a91e13e655245fac9quot;/ETag /CopyObjectResult [..] I have now been able to get the upgrade to succeed (including a subsequent verify with --data) by patching line 359 of backends/s3c.py, to read: if root.tag == 'CopyObjectResult': The S3-compatible backend in this case is StorageQloud ( https://www.greenqloud.com/storageqloud/ ). Does this look like a deviation from S3-compatibility, which I should report to GreenQloud? Yes. They are not declaring the proper XML namespace (which would be http://s3.amazonaws.com/doc/2006-03-01/) for the CopyObjectResult tag. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Problems after upgrading from 2.7 to 2.10
On 08/24/2014 07:25 AM, Dan Johansson wrote: Hi All, After upgrading to 2.10.1 I encounter the following error each time I try to do anything (fsck, s3qladm upgrade): # /usr/bin/fsck.s3ql --no-ssl --batch s3://abcdef.dmj.nu Using CONNECT proxy proxy.dmj.nu:8080 Encountered ConnectionError exception (Tunnel connection failed: 403 Forbidden), retrying call to Backend.open_read for the 3-th time... Encountered ConnectionError exception (Tunnel connection failed: 403 Forbidden), retrying call to Backend.open_read for the 4-th time... Running the same command from a second machine (not yet upgraded) works fine: # /usr/bin/fsck.s3ql --no-ssl --batch s3://abcdef.dmj.nu Starting fsck of s3://test.dmj.nu Using cached metadata. Checking DB integrity... Creating temporary extra indices... --- 8 --- Completed fsck of s3://abcdef.dmj.nu Both machines are in the same network and use the same proxy. What's happening? Looks like you can't connect to the proxy from the first machine. Are you sure you are using this proxy on the second machine? Note the conspicuous absence of the Using CONNECT proxy proxy.dmj.nu:8080 in your second quoted output. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Problems after upgrading from 2.7 to 2.10
On 08/25/2014 12:58 PM, Dan Johansson wrote: On 24.08.2014 23:37, Nikolaus Rath wrote: On 08/24/2014 07:25 AM, Dan Johansson wrote: Hi All, After upgrading to 2.10.1 I encounter the following error each time I try to do anything (fsck, s3qladm upgrade): # /usr/bin/fsck.s3ql --no-ssl --batch s3://abcdef.dmj.nu Using CONNECT proxy proxy.dmj.nu:8080 Encountered ConnectionError exception (Tunnel connection failed: 403 Forbidden), retrying call to Backend.open_read for the 3-th time... Encountered ConnectionError exception (Tunnel connection failed: 403 Forbidden), retrying call to Backend.open_read for the 4-th time... Running the same command from a second machine (not yet upgraded) works fine: # /usr/bin/fsck.s3ql --no-ssl --batch s3://abcdef.dmj.nu Starting fsck of s3://test.dmj.nu Using cached metadata. Checking DB integrity... Creating temporary extra indices... --- 8 --- Completed fsck of s3://abcdef.dmj.nu Both machines are in the same network and use the same proxy. What's happening? Looks like you can't connect to the proxy from the first machine. Are you sure you are using this proxy on the second machine? Note the conspicuous absence of the Using CONNECT proxy proxy.dmj.nu:8080 in your second quoted output. I can connect to the proxy allright: # wget www.google.com --2014-08-25 21:54:29-- http://www.google.com/ Resolving proxy.dmj.nu (proxy.dmj.nu)... 192.168.1.1 Connecting to proxy.dmj.nu (proxy.dmj.nu)|192.168.1.1|:8080... connected. [...] Not relevant, sorry. wget doesn't use CONNECT proxying. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] s3ql crash by read timeout of swift
On 08/27/2014 03:25 AM, motomura wrote: I use s3ql-2.10.1 in with OpenStack Swift (/v1) as backed storage. s3ql crashed. Below logs (swift.log and mount.log) when s3ql is crashed. May be the reason of s3ql'crash is read timeout of swift? Can I avoid this problem by setting of s3ql or swift? s3ql server's ip address is 192.168.21.66. [...] File /usr/lib64/python3.3/site-packages/s3ql-2.10.1-py3.3-linux-x86_64.egg/s3ql/backends/swift.py, line 232, in _do_request raise HTTPError(resp.status, resp.reason, resp.headers) s3ql.backends.s3c.HTTPError: 408 Request Timeout The problem is that the swift server returned a 408 HTTP code, and S3QL doesn't know what to do with it. Probably retrying would be the right choice, but to be sure I'd really need to have some documentation of the status codes and their meaning that Swift may return. For example, S3 has a comprehensive listing of status codes and the expected response of the client at http://docs.aws.amazon.com/AmazonS3/latest/API/ErrorResponses.html. But there is nothing comparable in the Swift documentation at http://docs.openstack.org/api/openstack-object-storage/1.0/content/PUT_createOrReplaceObject__v1__account___container___object__storage_object_services.html. This is why, in general terms, S3QL's Google Storage and Amazon S3 support is more robust than OpenStack/swift support. If you report this issue at https://bitbucket.org/nikratio/s3ql/issues I can add support for 408 error to S3QL. However, in order to fix this properly it'd be great if you could contact your storage provider and ask them if they have information about the error codes that the swift server may return, and how the client is supposed to react to them. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
[s3ql] [ANNOUNCE] S3QL 1.19 has been released
Dear all, I am pleased to announce a new maintenance release of S3QL, version 1.19. Please note that this is only a maintenance release. Development of S3QL takes place in the 2.x series. The 1.x releases receive only selected bugfixes and are only maintained for older systems that do not support Python 3.3. For systems with Python 3.3 support, using the most recent S3QL 2.x version is strongly recommended. From the changelog: 2014-08-25, S3QL 1.19 * SECURITY UPDATE. Fixed a remote code execution vulnerability. For non-encrypted file systems, an attacker with control over the communication with the storage backend or the ability to manipulate the data stored in the backend was able to trigger execution of arbitrary code by mount.s3ql. Encrypted file systems were protected against this if the attacker did not know the file system passphrase. Mounting an encrypted file system prepared by an attacker (which is possible if the attacker shares the file system passphrase) thus allowed the attacker to execute arbitrary code even when using encryption. Please report any bugs on the mailing list (s3ql@googlegroups.com) or the issue tracker (https://bitbucket.org/nikratio/s3ql/issues). Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Unexpected server reply to copy operation (upgrading from 20 to 21)
On 09/01/2014 07:54 AM, Rich B wrote: On Thursday, August 14, 2014 9:02:40 PM UTC-4, Nikolaus Rath wrote: On 08/14/2014 03:17 PM, Adam Watkins wrote: Upgrading from revision 20 to 21... Unexpected server reply to copy operation: 200 OK Date: Thu, 14 Aug 2014 21:04:15 + Server: RestServer/1.0 Content-Length: 189 Content-Type: application/xml ETag: d7ed6a310a47b69a91e13e655245fac9 Cache-Control: no-cache Connection: close ?xml version=1.0 encoding=UTF-8? CopyObjectResultLastModified2014-08-14T21:04:15.000Z/LastModified ETagquot;d7ed6a310a47b69a91e13e655245fac9quot;/ETag /CopyObjectResult [...] Does this look like a deviation from S3-compatibility, which I should report to GreenQloud? Yes. They are not declaring the proper XML namespace (which would be http://s3.amazonaws.com/doc/2006-03-01/ http://www.google.com/url?q=http%3A%2F%2Fs3.amazonaws.com%2Fdoc%2F2006-03-01%2Fsa=Dsntz=1usg=AFQjCNHS3kxugwRdqhZ6oS1Bzav7PPtrmQ) for the CopyObjectResult tag. Best, -Nikolaus Sorry to re-open an old thread, But I've run into a similar problem with Dream Host's S3 Compatible DreamObjects. When I create a filesystem, touch a file, and then umount it I get the following: [...] Backing up old metadata... Unexpected server reply: expected XML, got: 200 OK Date: Mon, 01 Sep 2014 14:45:08 + Server: Apache Transfer-Encoding: chunked Content-Type: binary/octet-stream Is this the same problem? Almost. The Dreamhost response is different from the GreenQloud response, but they're both not S3-like. If so, I'd like to report the bug to Dream Host, but the link Nikolaus provided above does not work, so I don't know how to describe the bug to them. That wasn't a link, http://s3.amazonaws.com/doc/2006-03-01/; is the XML namespace (which happens to look like an URL), that was not correctly declared by GreenQloud. Could some one provide a link to (of description of) the S3 API call that I piont to when submitting a bug report to D H? Here's an example for a proper response: 200 OK Date: Mon, 01 Sep 2014 14:45:08 + Server: Apache Transfer-Encoding: chunked Content-Type: text/xml; charset=utf-8 ?xml version=1.0 encoding=UTF-8? CopyObjectResult xmlns=http://s3.amazonaws.com/doc/2006-03-01/; LastModified2009-10-28T22:32:00/LastModified ETag9b2cf535f27731c974343645a3985328/ETag /CopyObjectResult Note the difference in Content-Type, and the presence of a proper response body. You can find more information at This from http://docs.aws.amazon.com/AmazonS3/latest/API/RESTObjectCOPY.html, but note that the examples at the end are actually incomplete (or outdated) and do not correspond to what S3 is actually returning. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] mount.s3ql hangs on version 2.11
On 09/03/2014 01:25 PM, Ido wrote: Hello, I've been using s3ql for about a year now, and I'm very happy / excited about it! Unfortunately, the recent upgrade seems to hang it on my system. This is what I'm getting: sudo mount.s3ql --cachesize=500 --authfile=[...] [s3_dest] [mount_root] Using 8 upload threads. Autodetected 499944 file descriptors available for cache entries Using cached metadata. Mounting filesystem... And then it just hangs there... Any ideas? This was a deliberate change. From the changelog: * mount.s3ql no longer daemonizes on its own. With a modern init system this should no longer be necessary, and when running mount.s3ql from the command line the shell can be used to put the process into background. I have since learned that many people want this feature though, so this change will be reverted in the next release. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Pros/Cons of using --nfs option
On 09/04/2014 07:31 AM, Nick Carboni wrote: I am using s3ql for an application that may require the user to read large amounts of data over an NFS (or CIFS) export of the s3ql file system. I came across the --nfs option for mount.s3ql which is described as enabling some optimizations for exporting the file system over NFS. I'm assuming there is some downside to specifying this option as the default is false. Is there any documentation on what this option does and how it could hurt or help performance? It increases the size of the local metadata database a bit, and it may reduce overall performance of metadata operations a bit (because S3QL has to keep an additional index up-to-date). Conversely, it should significantly increase the performance of some NFS requests, but I don't know how frequent this specific operation is (lookup of . or .. by name). As far as I know, no one has yet measured how big any of these effects actually are. Best, -Niko -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Re: Unexpected server reply to copy operation (upgrading from 20 to 21)
On 09/09/2014 12:20 PM, Adam Watkins wrote: Regarding my original issue with StorageQloud, GreenQloud have now responded after 25 days with the following: Part of our object storage solution involves third party software, so we will have to ask them to fix the namespace declaration. Unfortunately this will not going to happen soon. but probably will be fixed when we upgrade the version. If GreenQloud can guarantee that copies always succeed if the status is 200 (in contrast to S3, where error may only be indicated in the response body), this will probably help you: https://bitbucket.org/nikratio/s3ql/issue/86/add-copy-always-succeeds-backend-option Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] clone_fs.py ChecksumError: Invalid metadata
On 09/22/2014 12:17 PM, Nick Carboni wrote: Here is everything I've done. The commands are actually entered from a C program using fork(2) and execl(3), but should be functionally equivalent to running them from the command line which I've done below: $ mkfs.s3ql local:///mnt/bds/7625eec6-843c-4a4d-80fe-e0718dbab53b [...] I also tried the clone_fs.py command without the folder name on the target location. I'm running version 1.19 on Ubuntu 12.04 : I'm baffled. Could you try to run my sequence of commands (including the wget download)? Thanks, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] clone_fs.py ChecksumError: Invalid metadata
On 09/23/2014 06:16 AM, Nick Carboni wrote: I did some additional testing and eventually got it to work with other directories. When I create the backend mountpoint through my C program I drop some metadata in the directory in a file named meta and a lost+found directory also gets created. After removing these the clone_fs.py script ran just fine. Is it possible that either of those could be what was tripping up the clone script? Yes. Don't do that. The directory you pass to s3ql is for s3ql's exclusive use, no other files allowed. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Broken pipe error
Hi Brian, On 09/30/2014 06:36 AM, Brian Pribis wrote: Running tests for s3ql-2.11.1 on test: test_put_s3error_med[plain-mock_gs] I get the following error: Traceback (most recent call last): File /usr/local/lib/python3.4/dist-packages/dugong-3.0-py3.4.egg/dugong/__init__.py, line 557, in _co_send len_ = self._sock.send(buf) BrokenPipeError: [Errno 32] Broken pipe Thanks for reporting. Upgrading to python-dugong 3.3 should fix this problem. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Unclear on zero deduplication and cache for local backend
p6su5t8...@snkmail.com writes: Hi, I have recently tried out s3ql on Debian testing, and I have a few questions. I'm using s3ql with local storage, without encryption or compression. I set threads to 1 as a baseline [...] . I find when I specify cachesize manually to be small or zero that my write throughput goes down by several orders of magnitude. Is using no cache unsupported? Yes, this is not supported. You are right that if the backend storage is a local disk, this could be made to work. However, S3QL was designed for network storage, and the local storage backend was added for use with a network file system (like sshfs) and testing, and not as an efficient method to utilize your local disk. In theory, there are several optimizations one could implement with the local backend (not requiring a cache being one of them). However, I don't think this is worth it. I don't think that even with additional optimizations, there'd be little reason not to use e.g. dm-crypt with btrfs to get very similar features with orders of magnitude better performance. I don't mind a small performance loss but when I use a zero cache size I get throughput of around 50 kilobytes per second, which suggests that I'm running up against an unexpected code path. Read performance is okay even in that case. I think with zero cache, S3QL probably downloads, updates, uploads and removes a cache entry for every single read() call. The next thing I'm wondering a lot about is the deduplication. In my test, I'm writing all zeroes. I write a megabyte using one block of a 1MB block size using dd, and then I write 1024 blocks of a kilobyte each. I then also write 2MB or 4MB at a time. I'd expect that deduplication would catch these very trivial cases and that I'd only see one entry of at most 2^n bytes, where 2^n represents the approximate block size of the deduplication. Yes, this is what should happen. I'd also expect 2^n to be smaller than a megabyte (maybe like a single 64k block). That's probably not the case. S3QL de-duplicates on the level of storage objects. You specify the maximum storage object size at mkfs.s3ql time with the --blocksize option, and the default is 10 MB. To see de-duplication in action, you either need to write more data, or you need to write smaller, but identical files: $ echo hello, world foo $ echo hello, world bar ..in this case s3ql will store only one storage object (containing hello, world) in the backend. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] S3QL backup restore
On 10/07/2014 04:41 AM, Sven Martin wrote: I have an Amazon S3 bucket mounted on an Ubuntu 14.04 server, which serves as my ownCloud server (EFSS). This bucket is configured as the data location for ownCloud. So all the users' files end up here. What is the best way to: - backup this mounted S3 bucket I'd suggest following the steps described at http://www.rath.org/s3ql-docs/contrib.html#s3-backup-sh. - restore this backup in the event of data loss You can just mount the file system and copy everything back using cp, rsync or (for slightly better performance), pcp.py (http://www.rath.org/s3ql-docs/contrib.html#pcp-py). Remark: I have installed S3QL using these instructions https://bitbucket.org/nikratio/s3ql/wiki/installation_ubuntu. I do not see the /contrib folder anywhere. Did you look at /usr/share/doc/s3ql/README.Debian? Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] S3QL backup restore
Sven Martin sproeg...@gmail.com writes: My only concern is: Will I allways be able to mount the s3ql file system? Can this get corrupted somehow? I'm pretty sure there still a number of bugs in S3QL. And even if there were none, there's always the danger of broken hardware on your computer or the remote backend. So yeah, things can get corrupted. That said, there a number of people using S3QL regularly without having their file system corrupted. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] I need some help S3 + s3ql +ec2
Raquel Carrillo raquel.carrillo.solorz...@gmail.com writes: When list (ls) the contents of the mounted directory, using de commandline in ec2, all that shows to me is lost+found. But When I see the content of the bucket using the web console in s3 I see files like: s3ql_metadata s3ql_seq_no_1 Why can't I see these files on my mounted directory? and what do they represent? There is no 1:1 mapping between files in S3QL and storage objects in S3. A single file can be spread between multiple objects, and an object can hold data for multiple files. You should treat the contents of the bucket as a black-box. To give an analogy: if you open a .jpg file in a text editor, you wouldn't expect to see the picture (even if it contained some text). So if you open an S3QL file system in some generic S3 browser, you're not going to see your file system contents. 2. I made a new directory, using commandline in the s3ql directory with: sudo mkdir test-dir sudo touch testfile.txt But I can't see these files in the web console of my bucket in s3. Same reason. 3- This one is conceptual if for example I need to save my_photo.jpg in the bucket and I use s3ql, the file will be saved, and I will be able to see it with the web console like my_photo.jpg or it will be saved in another format? You should be able to answer that question yourself now :-). Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] S3QL on local storage backend provided by NFS - suspected incompatibility?
Super Strobi strobis...@gmail.com writes: **In summary**: is this setup supported by S3QL: local storage backend by NFS share? Yes, that should work. Not sure why you're having performance issues. I'd try to isolate the problem. For example, does the same thing happen if you use the local backend? Or use the NFS backend, but don't export the mountpoint via CIFS or NFS? Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] S3QL on local storage backend provided by NFS - suspected incompatibility?
On 10/18/2014 12:00 AM, Super Strobi wrote: PS. as per documentation uncertainty: if the local ~/.s3ql is lost, is the s3ql filesystem still mountable? If the file system was unmounted properly, yes. Otherwise you may loose all data that was modified since the last metadata upload. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] S3QL on local storage backend provided by NFS - suspected incompatibility?
On 10/18/2014 01:48 PM, Super Strobi wrote: [root@strobiserver .s3ql]# cat /etc/mtab [...] local:///mnt/local-nfs/myfsdata-ttt /mnt/s3ql-ttt-local fuse.s3ql rw,nosuid,nodev,relatime,user_id=0,group_id=0,default_permissions,allow_other 0 0 [...] ls -alR /mnt/s3ql-ttt-local does show the content, but files cannot be opened (hangs) ls -alR /mnt/local-nfs hangs too. The former is caused by the latter. If requests to the backend directory hang, S3QL is bound to hang as well once the cache is full. Fix your NFS mount, and the S3QL mountpoint will become accessible again as well. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] S3QL on local storage backend provided by NFS - suspected incompatibility?
On 10/19/2014 05:50 AM, Super Strobi wrote: Dumping metadata... ..objects.. ..blocks.. Uncaught top-level exception: Traceback (most recent call last): File /bin/fsck.s3ql, line 9, in module load_entry_point('s3ql==2.9', 'console_scripts', 'fsck.s3ql')() File /usr/lib/python3.3/site-packages/s3ql/fsck.py, line 1217, in main dump_metadata(db, fh) File /usr/lib/python3.3/site-packages/s3ql/metadata.py, line 137, in dump_metadata dump_table(table, order, columns, db=db, fh=fh) File deltadump.pyx, line 317, in s3ql.deltadump.dump_table (src/s3ql/deltadump.c:4096) File deltadump.pyx, line 364, in s3ql.deltadump.dump_table (src/s3ql/deltadump.c:3746) ValueError: Can't dump NULL values This looks like a bug. Could you report it athttps://bitbucket.org/nikratio/s3ql/issues? Thanks! Mounting is not possible: [root@strobiserver ~]# mount.s3ql local:///mnt/local-nfs/myfsdata-ttt /mnt/s3ql-ttt-local --allow-other --nfs Using 4 upload threads. Autodetected 4052 file descriptors available for cache entries Enter file system encryption passphrase: Using cached metadata. File system damaged or not unmounted cleanly, run fsck! Rerunning fsck gives same output as above. Suggestions to get out of this? Does the file system metadata (i.e., file names, sizes, permissions) contain confidential information? If not, can you upload it somewhere and send me the link? Then I'll take a look. Otherwise I can give you some commands to execute, but it'll take several emails back and forth... Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Cache is not being uploaded to the backend
On 10/19/2014 08:00 AM, Rich B wrote: When I run s3qlstat it will sometimes hang and sometimes crash: $ s3qlstat --debug /mnt/MyBucket/ 2014-10-19 10:08:09.369 4344 MainThread root.excepthook: Uncaught top-level exception: Traceback (most recent call last): File /usr/bin/s3qlstat, line 9, in module load_entry_point('s3ql==2.11.1', 'console_scripts', 's3qlstat')() File /usr/lib/s3ql/s3ql/statfs.py, line 49, in main buf = llfuse.getxattr(ctrlfile, 's3qlstat', size_guess=256) File fuse_api.pxi, line 206, in llfuse.capi.getxattr (src/llfuse/capi_linux.c:21045) OSError: [Errno 5] Input/output error: '/mnt/MyBucket/.__s3ql__ctrl__' This is probably not related to your other problems, but it might be easier to debug. What does ~/.s3ql/mount.log and ~/.s3ql/mount.s3ql_crit.log contain after such a crash? Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Cache is not being uploaded to the backend
On 10/19/2014 08:00 AM, Rich B wrote: Hello, I'm trying to use S3QL with Google Storage, but I'm consistently running into a problem where blocks from the cache are not making it to the backend. I'm running on Ubuntu 14.04 and using S3QL from the PPA (version 2.11.1+dfsg-1~trusty1). My mount command looks like this: $ mount.s3ql --threads 4 --metadata-upload-interval 7200 --allow-other --debug --cachedir /data/s3ql-cache --cachesize 1048567 gs://MyBucket/MyPrefix /mnt/MyBucket Here is a sample from the mount log sometime after copying 200+ files (0.5-3MB each) into the mount point: 2014-10-19 09:57:42.902 28314:Dummy-23 (name)s.statfs: statfs(): start 2014-10-19 09:57:44.365 28314:CommitThread (name)s.put: timeout, returning 2014-10-19 09:57:44.365 28314:CommitThread (name)s.put: waiting for reader.. 2014-10-19 09:57:49.382 28314:CommitThread (name)s.put: timeout, returning 2014-10-19 09:57:49.382 28314:CommitThread (name)s.put: waiting for reader.. 2014-10-19 09:57:52.902 28314:Dummy-20 (name)s.statfs: statfs(): start 2014-10-19 09:57:53.237 28314:Thread-6 (name)s.close: ObjectW(s3ql_data_4).close(): start 2014-10-19 09:57:53.238 28314:Thread-6 (name)s._do_request: preparing PUT /MyPrefixs3ql_data_4?None, qs=None 2014-10-19 09:57:53.239 28314:Thread-6 (name)s._send_request: _send_request(): PUT /MyPrefixs3ql_data_4 2014-10-19 09:57:53.351 28314:Thread-5 (name)s.close: ObjectW(s3ql_data_3).close(): start 2014-10-19 09:57:53.352 28314:Thread-5 (name)s._do_request: preparing PUT /MyPrefixs3ql_data_3?None, qs=None 2014-10-19 09:57:53.353 28314:Thread-5 (name)s._send_request: _send_request(): PUT /MyPrefixs3ql_data_3 2014-10-19 09:57:54.383 28314:CommitThread (name)s.put: timeout, returning 2014-10-19 09:57:54.383 28314:CommitThread (name)s.put: waiting for reader.. 2014-10-19 09:57:54.702 28314:Thread-3 (name)s.close: ObjectW(s3ql_data_1).close(): start 2014-10-19 09:57:54.702 28314:Thread-3 (name)s._do_request: preparing PUT /MyPrefixs3ql_data_1?None, qs=None 2014-10-19 09:57:54.703 28314:Thread-3 (name)s._send_request: _send_request(): PUT /MyPrefixs3ql_data_1 2014-10-19 09:57:59.384 28314:CommitThread (name)s.put: timeout, returning 2014-10-19 09:57:59.384 28314:CommitThread (name)s.put: waiting for reader.. 2014-10-19 09:57:59.793 28314:Thread-4 (name)s.close: ObjectW(s3ql_data_2).close(): start 2014-10-19 09:57:59.794 28314:Thread-4 (name)s._do_request: preparing PUT /MyPrefixs3ql_data_2?None, qs=None 2014-10-19 09:57:59.795 28314:Thread-4 (name)s._send_request: _send_request(): PUT /MyPrefixs3ql_data_2 2014-10-19 09:58:02.902 28314:Dummy-24 (name)s.statfs: statfs(): start 2014-10-19 09:58:03.506 28314:Thread-6 (name)s.wrapped: Encountered ConnectionTimedOut exception (send/recv timeout exceeded), retrying call to ObjectW.close for the 33-th time... 2014-10-19 09:58:03.631 28314:Thread-5 (name)s.wrapped: Encountered ConnectionTimedOut exception (send/recv timeout exceeded), retrying call to ObjectW.close for the 33-th time... 2014-10-19 09:58:04.385 28314:CommitThread (name)s.put: timeout, returning 2014-10-19 09:58:04.385 28314:CommitThread (name)s.put: waiting for reader.. 2014-10-19 09:58:05.282 28314:Thread-3 (name)s.wrapped: Encountered ConnectionTimedOut exception (send/recv timeout exceeded), retrying call to ObjectW.close for the 33-th time... 2014-10-19 09:58:09.386 28314:CommitThread (name)s.put: timeout, returning 2014-10-19 09:58:09.386 28314:CommitThread (name)s.put: waiting for reader.. 2014-10-19 09:58:10.943 28314:Thread-4 (name)s.wrapped: Encountered ConnectionTimedOut exception (send/recv timeout exceeded), retrying call to ObjectW.close for the 33-th time... 2014-10-19 09:58:12.902 28314:Dummy-19 (name)s.statfs: statfs(): start 2014-10-19 09:58:14.387 28314:CommitThread (name)s.put: timeout, returning 2014-10-19 09:58:14.387 28314:CommitThread (name)s.put: waiting for reader.. 2014-10-19 09:58:19.388 28314:CommitThread (name)s.put: timeout, returning 2014-10-19 09:58:19.388 28314:CommitThread (name)s.put: waiting for reader.. 2014-10-19 09:58:22.902 28314:Dummy-18 (name)s.statfs: statfs(): start 2014-10-19 09:58:24.389 28314:CommitThread (name)s.put: timeout, returning 2014-10-19 09:58:24.389 28314:CommitThread (name)s.put: waiting for reader.. Could you give some more context? In particular, I'd like to see the logs from the 32st and 1st retry. If I manually kill mount.s3ql and umount the filesystem, then run fsck.s3ql, the cache does get uploaded (with some difficulty): $ fsck.s3ql --cachedir /data/s3ql-cache gs://MyBucket/MyPrefix Starting fsck of gs://MyBucket/MyPrefix Using cached metadata. Remote metadata is outdated. Checking DB integrity... Creating temporary extra indices... Checking lost+found... Checking cached objects... Committing block 0 of inode 143 to backend Committing block 0 of inode 289 to backend
Re: [s3ql] S3QL on local storage backend provided by NFS - suspected incompatibility?
On 10/19/2014 11:05 AM, Super Strobi wrote: Does the file system metadata (i.e., file names, sizes, permissions) contain confidential information? If not, can you upload it somewhere and send me the link? Then I'll take a look. I can share if it, however I run into this: [root@strobiserver local-nfs]# s3qladm download-metadata local:///mnt/local-nfs/myfsdata-ttt/ [...] Sorry, I should have been more clear. Just upload the .db file from ~/.s3ql/. No need to download anything. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] benchmark.py AttributeError: 'ArgumentParser' object has no attribute 'add_ssl'
good4...@gmail.com writes: Debian Wheezy. Python 3.3.0 running in pyvenv. (virtualenv-3.3.0) nick@host:~/build/s3ql-2.11/contrib$ python -V Python 3.3.0 (virtualenv-3.3.0) nick@host:~/build/s3ql-2.11/contrib$ python benchmark.py Traceback (most recent call last): File benchmark.py, line 225, in module main(sys.argv[1:]) File benchmark.py, line 71, in main options = parse_args(args) File benchmark.py, line 55, in parse_args parser.add_ssl() AttributeError: 'ArgumentParser' object has no attribute 'add_ssl' Can anyone help? If you make a small change to benchmark.py it should work: diff --git a/contrib/benchmark.py b/contrib/benchmark.py --- a/contrib/benchmark.py +++ b/contrib/benchmark.py @@ -52,7 +52,7 @@ parser.add_authfile() parser.add_quiet() parser.add_debug() -parser.add_ssl() +parser.add_backend_options() parser.add_version() parser.add_storage_url() parser.add_argument('file', metavar='file', type=argparse.FileType(mode='rb'), Thansk for the report, I'll fix this in the next release. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Cache is not being uploaded to the backend
Rich B rich...@gmail.com writes: On Sunday, October 19, 2014 1:45:54 PM UTC-4, Nikolaus Rath wrote: On 10/19/2014 08:00 AM, Rich B wrote: When I run s3qlstat it will sometimes hang and sometimes crash: This is probably not related to your other problems, but it might be easier to debug. What does ~/.s3ql/mount.log and ~/.s3ql/mount.s3ql_crit.log contain after such a crash? It took a couple tries to recreate the problem: $ date;s3qlstat /mnt/gl-s3ql Sun Oct 19 15:26:44 EDT 2014 Uncaught top-level exception: Traceback (most recent call last): File /usr/bin/s3qlstat, line 9, in module load_entry_point('s3ql==2.11.1', 'console_scripts', 's3qlstat')() File /usr/lib/s3ql/s3ql/statfs.py, line 49, in main buf = llfuse.getxattr(ctrlfile, 's3qlstat', size_guess=256) File fuse_api.pxi, line 206, in llfuse.capi.getxattr (src/llfuse/capi_linux.c:21045) OSError: [Errno 5] Input/output error: '/mnt/gl-s3ql/.__s3ql__ctrl__' mount.log: [...] 2014-10-19 15:26:44.990 18447:Dummy-18 (name)s.getxattr: getxattr(2, b's3qlstat'): start 2014-10-19 15:26:44.991 18447:Dummy-18 (name)s.extstat: extstat(%d): start [ Nothing more from Dummy-18 ] I'm afraid at the moment it beats me completely how this can possibly happen. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Cache is not being uploaded to the backend
Rich B rich...@gmail.com writes: Could you give some more context? In particular, I'd like to see the logs from the 32st and 1st retry. Because of log rotations, I had to re-run the test to get the 1st and 32nd retry. This time more of the blocks managed to get uploaded. Some of the threads seemed to have better luck than others: [...] If I manually kill mount.s3ql and umount the filesystem, then run fsck.s3ql, the cache does get uploaded (with some difficulty): $ fsck.s3ql --cachedir /data/s3ql-cache gs://MyBucket/MyPrefix Starting fsck of gs://MyBucket/MyPrefix Using cached metadata. Remote metadata is outdated. Checking DB integrity... Creating temporary extra indices... Checking lost+found... Checking cached objects... Committing block 0 of inode 143 to backend Committing block 0 of inode 289 to backend Committing block 0 of inode 212 to backend Committing block 0 of inode 338 to backend Committing block 0 of inode 334 to backend Committing block 0 of inode 154 to backend Committing block 0 of inode 350 to backend Committing block 0 of inode 354 to backend Committing block 0 of inode 224 to backend Committing block 0 of inode 215 to backend Committing block 0 of inode 221 to backend Committing block 0 of inode 339 to backend Committing block 0 of inode 278 to backend Committing block 0 of inode 178 to backend Encountered ConnectionTimedOut exception (send/recv timeout exceeded), retrying call to ObjectW.close for the 3-th time... Encountered ConnectionTimedOut exception (send/recv timeout exceeded), retrying call to ObjectW.close for the 4-th time... Encountered ConnectionTimedOut exception (send/recv timeout exceeded), retrying call to ObjectW.close for the 5-th time... Encountered ConnectionTimedOut exception (send/recv timeout exceeded), retrying call to ObjectW.close for the 6-th time... Encountered ConnectionTimedOut exception (send/recv timeout exceeded), retrying call to ObjectW.close for the 7-th time... Encountered ConnectionTimedOut exception (send/recv timeout exceeded), [...] To me this looks just like a crappy network connection. You could try to increase the send/recv timeout (using the --backend-options argument, see manual). I'm stumped. Has anyone got an idea of what might be causing my problem? Do you have the same problem when you store data in Amazon S3 from the same system? Yes, I just tried Amazon and I get similar results. This is consistent with a bad connection. If you have the necessary SSL-fu, you could also do a traffic capture using Wireshark. That's gonna be a bit tricky, but it should tell us a lot more. I've worked with tcpdump/wireshark before, but my SSL-fu is weak can you tell me what I should look for? I'm afraid not, I never got it to work myself. But you're lucky, if you have the same problem with S3, just use --backend-options nossl, and capture that traffic. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Benchmarks and Clustering.
gi...@liquidhealthcare.com writes: Hi Guys I've been looking through the wiki and I've noticed that this file system doesn't support mounting on multiple machines, I wanted to check whether there has been any benchmarks with this software. I don't think anything has been published. You can always run benchmark.py on your system to get some rough numbers. How many do we know how many uploads this can take before it maxes out. I don't understand the question. There's no limit on uploads. Do you mean the total number of files? That should scale n * log(n), and I have tested with around ~30 million files. You can easily go above that, but the database size scales with the number of files. We'd normally be looking for something which supports clustering and I wanted to ask what would be involved in making this support clustering? What do you mean with clustering? Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] ghost folder on nfs client
On 10/21/2014 09:33 PM, Maxim Kostrikin wrote: Suddenly I found ghost folder on nfs client of s3ql. Its name misticaly doubles ls output On nfs server with s3ql no such behavior | user@server://nfs/sharedfs/logs$ ls -la ls: reading directory .: Too many levels of symbolic links total 0 drwxr-xr-x. 1 sfinx0 sfinx 0 Oct 21 23:23 . drwxr-xr-x. 1 root root 0 Oct 21 01:47 .. drwxr-xr-x. 1 sfinx0 sfinx 0 Oct 21 23:23 fenix drwxr-xr-x. 1 sfinx0 sfinx 0 Oct 21 23:23 fenix [...] on nfs server: mount.s3ql --cachedir /nfsexport/.s3cache/ --cachesize 33554432 --max-cache-entries 1048576 --threads 16 --nfs s3://bucker /nfsexport/sharedfs/ root@nfs:/nfsexport/sharedfs/logs# ls -la total 0 drwxr-xr-x 1 1100 1100 0 Oct 21 23:24 fenix root@nfs:/nfsexport/sharedfs/logs# ls * test | Thanks for the report! This looks like a bug in S3QL when being exported over NFS. Could you report it at https://bitbucket.org/nikratio/s3ql/issues? I'll look into it as soon as possible. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Cache is not being uploaded to the backend
On 10/22/2014 07:25 AM, Rich B wrote: In the mean time, I have some questions about the tcp-timeout backend option. The docs don't say what the units are, what the default value is, or how to use the option. From the code, it looks like the units are seconds, and the syntax is --backend-options tcp-timeout=integer, however when I try this syntax, python barfs: $ /usr/bin/mount.s3ql --debug --cachedir /data/s3ql-cache --cachesize 1048567 --backend-options no-ssl,tcp-timeout=12 s3://myBucket/myPrefix /mnt/myMount Using 8 upload threads. Autodetected 4040 file descriptors available for cache entries Uncaught top-level exception: Traceback (most recent call last): File /usr/bin/mount.s3ql, line 9, in module load_entry_point('s3ql==2.11.1', 'console_scripts', 'mount.s3ql')() File /usr/lib/s3ql/s3ql/mount.py, line 121, in main [...] return eval_coroutine(self.co_read_response(), self.timeout) File /usr/lib/python3/dist-packages/dugong/__init__.py, line 1361, in eval_coroutine if not next(crt).poll(timeout=timeout): File /usr/lib/python3/dist-packages/dugong/__init__.py, line 115, in poll return bool(poll.poll(timeout*1000)) # convert to ms TypeError: timeout must be an integer or None Am I using tcp-timeout incorrectly? What is the default timeout? Yes, this looks like a bug in S3QL. Could you report it at https://bitbucket.org/nikratio/s3ql/issues? Thanks! -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] S3QL backup restore
Hi Sven, When replying to emails on this list, please do not put your reply above the quoted text, and do not quote the entire message you're answering to. This makes it unnecessarily hard for other readers to understand the context of your email. Instead, please cut quoted parts that are not relevant to your reply, and insert your responses right after the points you're replying to (as I have done below). Thanks! Sven Martin sproeg...@gmail.com writes: Hi Nikolaus, The server I mounted a S3QL file system on is still in test. I installed owncloud and configured its data directory to be on a mounted S3 volume using S3QL. Rebooting automounts the volume. I was on holiday for a week and when I came back the system had no free disk space. It turned out the volume was not mounted and the mountpoint got filled up with owncloud data. I don't knwo what caused the volume to unmount. I deleted the data (user data, but it's test so no issue) to free up space, rebooted the server. Now I am unable to mount the S3 filesystem. It keeps saying Mounting filesystem... and does not return to the prompt. Running fsck.s3ql --force works but does not help. Any tips tricks? What S3QL version are you using? Note the following entries in the Changelog: 2014-09-04, S3QL 2.11.1 * By popular demand, mount.s3ql is now able to daemonize again and also does so by default. 2014-08-27, S3QL 2.11 * mount.s3ql no longer daemonizes on its own. With a modern init system this should no longer be necessary, and when running mount.s3ql from the command line the shell can be used to put the process into background. Best, Nikolaus PS: Please remember the first paragraph when replying. -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Does it make sense to rsync raw s3ql data?
Super Strobi strobis...@gmail.com writes: Hello, I'm currently investigating the possibility to synchronize 2 s3ql filesystems, but without mounting them. I assume you mean without mounting them at the same time? Otherwise I don't understand what you're doing. What I'm trying to achieve: - NAS1: contains the s3ql filesystem (which is s3ql-mounted by a linux server, through sshfs - but not 24/7) NAS1 is a computer, or a directory? What does it mean for the s3ql filesystem to be s3ql-mounted by a linux server? Is NAS1 that Linux server? Is sshfs the S3QL backend, or are you accessing S3QL via sshfs? - NAS2: contains a scp -pr of the NAS1 contents as a baseline. Of what exactly? The s3ql file system? Or the backend of the s3ql file system? I use the following command on NAS1 to rsync the delta's made by linux server, after which (off course) the s3ql filesystem got umounted properly - the delta should be around 2GB. I don't understand. What delta's? How are they made? 2. Is there a smarter way to synchronise *offline *s3ql file systems (so *without* having to mount them)? Yes, but the procedure depends on the backend. With the local backend, use rsync. With the remote storage backend, you need to use a tool specific to that service. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Does it make sense to rsync raw s3ql data?
On 11/05/2014 02:37 PM, Super Strobi wrote: Let me rephrase the question differently: If I watch the original myfsdata directory, and looking at nodes like: myfsdata/s3ql_data_/188/s3ql_data_188998 Is there any reason why rsync *thinks* they have changed with having added only 2GB of data inside the file system. I can't speak for rsync, but as far as S3QL is concerned, these files are write-once. In other words, they are created with their final content and may be deleted at some point, but in regular operation (i.e., excluding fsck.s3ql after a crash) they are never modified or recreated, no matter what you do inside the S3QL mountpoint. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
[s3ql] [ANNOUNCE] S3QL 2.12 has been released
Dear all, I am pleased to announce a new release of S3QL, version 2.12. From the changelog: 2014-11-09, S3QL tip * The s3c backend has a new 'dumb-copy' option. If this option is enabled, copy operations are assumed to have succeeded as soon as the server returns a '200 OK' status, without checking the contents of the response body. * The swift and s3c backends have a new 'disable-expect100' to disable support for the 'Expect: continue' header. Using this option allows S3QL to work with storage servers without proper HTTP 1.1 support, but may decrease performance as object data will be transmitted to the server more than once in some circumstances. * contrib/benchmark.py is now working again. * The `tcp-timeout` backend option is now actually working instead of resulting in a TypeError. * Fixed a problem where saving metadata would fail with ValueError: Can't dump NULL values. * s3qlstat now also gives information about cache usage. As usual, the release is available for download from https://bitbucket.org/nikratio/s3ql/downloads Please report any bugs on the mailing list (s3ql@googlegroups.com) or the issue tracker (https://bitbucket.org/nikratio/s3ql/issues). Starting with version 2.0, S3QL requires Python 3.3 or newer. For older systems, the S3QL 1.x branch (which only requires Python 2.7) will continue to be supported for the time being. However, development concentrates on S3QL 2.x while the 1.x branch only receives selected bugfixes. When possible, upgrading to S3QL 2.x is therefore strongly recommended. Enjoy, -Nikolaus -- Encrypted emails preferred. PGP fingerprint: 5B93 61F8 4EA2 E279 ABF6 02CF A9AD B7F8 AE4E 425C »Time flies like an arrow, fruit flies like a Banana.« signature.asc Description: OpenPGP digital signature
Re: [s3ql] S3QL 2.12 loops fsck
Hi Maxim, When replying to emails on this list, please do not put your reply above the quoted text, and do not quote the entire message you're answering to. This makes it unnecessarily hard for other readers to understand the context of your email. Instead, please cut quoted parts that are not relevant to your reply, and insert your responses right after the points you're replying to (as I have done below). Thanks! Maxim Kostrikin mkostri...@gmail.com writes: вторник, 11 ноября 2014 г., 1:13:49 UTC+6 пользователь Nikolaus Rath написал: Maxim Kostrikin mkost...@gmail.com javascript: writes: 2014-11-10 05:11:33.867 4539:MainThread s3ql.fsck.check: Creating temporary extra indices... 2014-11-10 05:11:35.765 4539:MainThread s3ql.fsck.check_lof: Checking lost+found... 2014-11-10 05:11:35.765 4539:MainThread s3ql.fsck.check_cache: Checking cached objects... 2014-11-10 05:11:39.559 4539:MainThread s3ql.fsck.check_names_refcount: Checking names (refcounts)... Last line stick for hours with no progress. fsck.s3ql eat on core 100% constantly. Please advise. Is this a new problem with 2.12, i.e. was it working with 2.11? What is the output if you press Ctrl-C? I hoped that 2.12 was the cause. 2.11.1 same result. Ctrl-C typing C^ Kill PID - interrupts the fsck process. You mean, you cannot abort the process with Ctrl-C? In that case, please apply the attached patch and send SIGUSR1 when it's hanging, this should give a proper stack trace. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout. --- a/src/s3ql/fsck.py +++ b/src/s3ql/fsck.py @@ -30,6 +30,7 @@ import textwrap import time import atexit +import faulthandler log = logging.getLogger(__name__) @@ -1139,6 +1140,9 @@ options = parse_args(args) setup_logging(options) +faulthandler.enable() +faulthandler.register(signal.SIGUSR1) + # Check if fs is mounted on this computer # This is not foolproof but should prevent common mistakes if is_mounted(options.storage_url):
[s3ql] [ANNOUNCE] New Dugong Version Released
Dear all, I have just released a new version of python-dugong (3.4). This release adds proper HTTP (i.e., non HTTPS) proxy support. If you would like to use this feature, you can update python-dugong and S3QL will use it automatically. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] mkfs/mount 403 Forbidden error
On 12/03/2014 10:50 AM, Andy Cress wrote: Using s3ql-1.17 and fuse-2.8.7-2.el6.x86_64 All other account/buckets on S3 work fine from the same system. Only this account/bucket gets this error, what might be wrong with it? This bucket gets the same error from multiple systems. Below is the sequence: mkfs 403 error clear download-metadata ok, empty mkfs works mount gets 403 error download-metadata error (with debug) I'm confused about what might be wrong with this particular bucket (or account). Is there something with AWS IAM that might cause this? That's quite possible, I never tried IAM. Other ideas? Disable IAM and see if it works? Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] s3ql permissions for read-only mount (without s3ql knowing that it is read-only)
Peter Schüller schuelle...@gmail.com writes: Is there a way to mount s3ql read-only by setting bucket permissions in a way that * mounting is still possible, but * changing the file system is not possible? If I disable all Put/Delete actions of S3 then mounting fails, probably because the fact that the FS is mounted is stored in some special file in the bucket. I can understand this, but what happens if I allow writing just to that file and not to the rest of the bucket? Is it even one file? Well, it's just one file, but a different on every mount. On every mount, S3QL creates a file s3ql_seq_no_XX, where XX is an increasing number, and deletes the n-th most recent s3ql_seq_no_YY file. If you can configure AWS to allow that, then it should work. To clarify: I do not want to mount in multiple locations, I just want to have a user that has no possibility to destroy the data but can still read it. That user will likely get warnings and/or crashes because of apparent file system corruption if he tries to access data that has been modified by the other active mount. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Connexion Cloudwatt
Hi netplus, When replying to emails on this list, please do not put your reply above the quoted text, and do not quote the entire message you're answering to. This makes it unnecessarily hard for other readers to understand the context of your email. Instead, please cut quoted parts that are not relevant to your reply, and insert your responses right after the points you're replying to (as I have done below). Thanks! netplus.r...@gmail.com writes: Hello, Thank you for your help. It seems that it's not the good URL. But the URL is not classic and I have got this message instead Invalid storage URL (because it's not well parse by python script I guess). This is an example of URL : swiftks://storage.fr1.cloudwatt.com/v1/AUTH_abcd1234efgh5678/region:container Like you see, the hostname is storage.fr1.cloudwatt.com/v1/AUTH_abcd1234efgh5678 and it's a problem. No, that's not the hostname. The hostname is storage.fr1.cloudwatt.com. You are not supposed to put the v1/... part into the storage URL, S3QL autodetects that. Best, -Nikolaus PS: Please remember the first paragraph when replying :-). -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Connection closed unexpectedly, fuse,s3ql
On 12/23/2014 04:17 AM, michael karotsieris wrote: Hello , I get and HTTPerror() and a connection closed unexpectedly error when i try to mount a s3ql container on a fuse fs in my machine .What happens next is that whenever i try to access the fuse fs ,the process i am using will go in uninterrubtible sleep and the only thing i can do is reboot.Does this sound like a known issue yet?Thanks in advance No, I haven't any any such reports before. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] increasing upload performance
On 01/25/2015 09:55 PM, Di Pe wrote: We have our local swift cluster and we have a few use cases for s3ql. One of them is keeping a mirror of a posix file system. For this we want to maximize write throughput per process. I am aware that each copy process is limited by the write throughput of the swift cluster which is between 50 and 60 MB/s. I thought that compression would be able to improve upload performance but no matter which compression type I use my 1GB test file always needs 24-28 sec to upload and rsync shows between 30 and 40 MB/s throughput. Is zlib perhaps too slow to do real on-the-fly compression and has to do most of the work after the file is uploaded? contrib/benchmark.py should be able to answer that question. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] OSError: [Errno 14] Bad address
Jeff Bogatay j...@bogatay.com writes: I am in the process of crafting a s3ql backed backup solution. During the development/testing I left the store mounted, installed some system updates and rebooted. Now I am unable to mount and/or check the store. Running 2.12 on ArchLinux. It has been several hours since I last wrote to the store. My last attempt was to delete the local metadata and have it rebuilt. Same error as below. Not sure what to do next or how to recover. Are these stores typically this fragile? What do you mean with the store? Are you talking about a remote storage server? In that case the fragility obviously depends on the server. Also, as a test I created a fresh mount, wrote to it, unmounted it, and remounted it without any issues. 2015-02-04 16:56:35.635 9617:MainThread s3ql.deltadump.dump_metadata: dump_table(ext_attributes): writing 0 rows [...] I am not sure what I'm looking at here. First you say it works, but then you quote an error message (and the formatting is pretty messed up). Can you be more precise as to when exactly the error occurs (and is it always the same)? Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Looking for best practices and some beginners questions: Using s3ql as backup target
Nikolaus Rath nikol...@rath.org writes: 4) How can I realize an additional backup of (the current state of) an s3ql file system e.g. to an external drive? Not easily. Instead I'd suggest to run an rsync job from your data directory to an S3QL mount point, and to your external drive. See https://bitbucket.org/nikratio/s3ql/wiki/FAQ#!is-there-a-way-to-make-the-cache-persistent-access-the-file-system-offline for details. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Anyone depend on API authentication for Google Storage?
Hi Andy, [ When replying to emails on this list, please do not put your reply above the quoted text, and do not quote the entire message you're answering to. This makes it unnecessarily hard for other readers to understand the context of your email. Instead, please cut quoted parts that are not relevant to your reply, and insert your responses right after the points you're replying to (as I have done below). Thanks! ] Andy Cress andycr...@gmail.com writes: On Sun, Feb 1, 2015 at 4:43 PM, Nikolaus Rath nikol...@rath.org wrote: Hello, Does anyone depend on using API / old-style authentication for accessing Google Storage with S3QL? I am considering to make OAuth2 the only supported authentication method for Google Storage. I believe that our usage could leverage whatever S3QL supports, but we do currently use the gs legacy API keys with s3ql v1.17 (i.e. https://code.google.com/apis/console/#project:467632721924:storage:legacy). The issue for us would be migrating existing users from the legacy API to OAuth2 without disrupting their access to a given bucket. If that is clearly documented and not too difficult, we could manage the transition. Thanks for your reply! This change would affect only S3QL 2.x. So the bigger disruption for you would probably be the upgrade of S3QL, not the change of the authentication method. As a matter of fact, you can use the legacy and OAuth2 API at the same time, so I think the transition should be very smooth. For s3ql usage, would OAuth2 change the tag from 'gs' to 'gs2', as was done with swift? I would be a little concerned if the logic changed for 'gs' to mean something else from version to version. No, there probably would not be a change of the tag. You can already use OAuth2 authentification with S3QL at the moment if you specify oauth2 as the login and your refresh token as the password. The change would (probably) be that S3QL just aborts if you specify anything but 'oauth2' as the login. Best, Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] s3ql.backends.s3c.S3Error: CannotPutData: (upgrading from 21 to 22)
Adam Watkins adam.gc.watk...@gmail.com writes: Encountered ConnectionClosed exception (connection closed unexpectedly), retrying call to Backend.lookup for the 3-th time... ..processed 147/16379 objects (0.9%, 0 bytes rewritten)..Uncaught top-level exception: Original/inner traceback (most recent call last): Traceback (most recent call last): File /usr/local/lib/python3.4/site-packages/s3ql-2.13-py3.4-linux-i686.egg/s3ql/common.py, line 482, in run Are you sure that you did not omit something here? This does not make sense, it should look sort-of like this: ..processed 147/16379 objects (0.9%, 0 bytes rewritten)..Uncaught top-level exception: Traceback (most recent call last): File /home/nikratio/in-progress/s3ql/bin/s3qladm, line 26, in module s3ql.adm.main(sys.argv[1:]) File /home/nikratio/in-progress/s3ql/src/s3ql/adm.py, line 94, in main return upgrade(options) File /home/nikratio/in-progress/s3ql/src/s3ql/common.py, line 549, in wrapper return fn(*a, **kw) File /home/nikratio/in-progress/s3ql/src/s3ql/adm.py, line 339, in upgrade update_obj_metadata(backend, backend_factory, db, options.threads) File /home/nikratio/in-progress/s3ql/src/s3ql/common.py, line 549, in wrapper return fn(*a, **kw) File /home/nikratio/in-progress/s3ql/src/s3ql/adm.py, line 452, in update_obj_metadata t.join_and_raise() File /home/nikratio/in-progress/s3ql/src/s3ql/common.py, line 503, in join_and_raise raise EmbeddedException(exc_info, self.name) s3ql.common.EmbeddedException: caused by an exception in thread Thread-19. Original/inner traceback (most recent call last): Traceback (most recent call last): File /home/nikratio/in-progress/s3ql/src/s3ql/common.py, line 482, in run self.run_protected() It'd be great if you could post the *full* traceback again (ideally also while preserving indentation and newlines, if your mailer insists on rewrapping you can attach the traceback as a separate file). Thanks, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] OSError: [Errno 14] Bad address
Jeff Bogatay j...@bogatay.com writes: I was running 1.0.2. I still had 1.0.1 in my archlinux pkg cache, so I downgraded and everything worked fine. Fsck completed without any issues. For anybody else out there experiencing this issue -- it's clearly an openssl problem. It'd be great if you could file an OpenSSL bug then. You probably don't want to stay with version 1.0.1 forever :-). Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] S3QL 2.13 upgrade broken (runtests.py fails on HostnameNotResolvable)
Strobi strobis...@gmail.com writes: S3QL needs at least dugong 3. Are you sure that you've got that installed? Try to run That should have been dugang 3.4, sorry. But it's correct in the user's guide :-P. # python3 -c 'import dugong; print(dugong.__version__)' # python3 -c 'import dugong; print(dugong.__version__)' 3.2 Well, there's your problem. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] fsck 2.12 hangs processing objects
On Feb 16 2015, Guilherme Barile guig...@gmail.com wrote: with high I/O scenarios in the past (using the 1.x series), where the mount daemon would die under high load, but managed to solve them all with fsck.s3ql after rebooting. Actually that doesn't sound like a solution at all. mount.s3ql should never crash, now matter how high the load is. Can you still reproduce that? If so, it'd be great if you could post the backtrace. This time it happened while I was performing 2 full backups with lots of small files, from my mounted s3qlfs to another s3 bucket via duplicity. I've experienced this when apache was under high load. When mount.s3ql stops, all other processes start to wait for io, increasing the load constantly, I can try to force this behaviour on another bucket after I restore this one. Please do! Btw, which version is it now? Your subject says 2.12. Filesystem was created on 2.11, I tried upgrading to 2.13 but couldn't due to db revision upgrade, so I had to compile 2.12 to try fsck/upgrade my volume - /usr/local/bin/fsck.s3ql --debug s3://my-bucket/home Now fsck.s3ql is hanging at ..processed 99500 objects so far.. fsck.log (with --debug) shows a lot of HEAD requests, It seems to be running a verify for every block (about 140). That should not happen. Please post more context for the logfile. What's the last message before the HEAD requests are starting? As this should not happen, I stopped fsck and ran it again to check the logs, the output doesn't show any errors https://gist.github.com/guigouz/7a6a624d97d12918b3f6 I think you may be running into a bug that has been fixed in S3QL 2.13. Could you try the attached patch? It backports the relevant change to 2.12. I also tried downloading a metadata backup, but it hangs at the same point, here's an excerpt of the ongoing log: What do you mean with at the same point? Downloading metadata should not issue any requests for data objects at all. fsck hangs at ..processed 99500 objects so far.., no matter which db backup I use Ah, so there is no problem with downloading the metadata backup. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout. diff --git a/src/s3ql/backends/s3c.py b/src/s3ql/backends/s3c.py --- a/src/s3ql/backends/s3c.py +++ b/src/s3ql/backends/s3c.py @@ -204,7 +204,7 @@ log.debug('list(%s, %s): start', prefix, start_after) keys_remaining = True -marker = start_after +marker = self.prefix + start_after prefix = self.prefix + prefix ns_p = self.xml_ns_prefix diff --git a/tests/mock_server.py b/tests/mock_server.py --- a/tests/mock_server.py +++ b/tests/mock_server.py @@ -222,7 +222,7 @@ 'IsTruncatedfalse/IsTruncated' ] count = 0 -for key in self.server.data: +for key in sorted(self.server.data): if not key.startswith(prefix): continue if marker and key = marker:
Re: [s3ql] fsck 2.12 hangs processing objects
On Feb 17 2015, Guilherme Barile guig...@gmail.com wrote: Have you seen s3ql running on similar setups ? Serving webpages / files directly ? No. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] fsck 2.12 hangs processing objects
On Feb 16 2015, Guilherme Barile guig...@gmail.com wrote: Hello, I use s3ql for quite some time as my apache webroot, which serves about 300gb of files and some php applications. I had few problems ^^^ you already said that. SCNR. with high I/O scenarios in the past (using the 1.x series), where the mount daemon would die under high load, but managed to solve them all with fsck.s3ql after rebooting. Actually that doesn't sound like a solution at all. mount.s3ql should never crash, now matter how high the load is. Can you still reproduce that? If so, it'd be great if you could post the backtrace. My current server was running s3ql 2.11 on ubuntu 14.04. Earlier today, I was performing full backups from that node to another s3 bucket using duplicity, and after lots of gbs transferred, s3ql failed and I had to reboot the server. Now fsck.s3ql is hanging at ..processed 99500 objects so far.. fsck.log (with --debug) shows a lot of HEAD requests, It seems to be running a verify for every block (about 140). That should not happen. Please post more context for the logfile. What's the last message before the HEAD requests are starting? I also tried downloading a metadata backup, but it hangs at the same point, here's an excerpt of the ongoing log: What do you mean with at the same point? Downloading metadata should not issue any requests for data objects at all. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] fsck 2.12 hangs processing objects
On Feb 16 2015, Guilherme Barile guig...@gmail.com wrote: Hello, I use s3ql for quite some time as my apache webroot, which serves about 300gb of files and some php applications. I had few problems with high I/O scenarios in the past (using the 1.x series), where the mount daemon would die under high load, but managed to solve them all with fsck.s3ql after rebooting. My current server was running s3ql 2.11 on ubuntu 14.04. Earlier [...] Btw, which version is it now? Your subject says 2.12. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] fsck 2.12 hangs processing objects
On Feb 18 2015, Guilherme Barile guig...@gmail.com wrote: Nikolaus, After fscking and upgrading my volume to the new db revision, I was able to mount the filesystem and access all the files without problems (even though I stress the filesystem a lot, I never lost data using s3ql). With the volume back online, I started to rsync about 280gb of files *from* S3QL to an EBS volume (everything running inside AWS, on the same region). I had 3 jobs in parallel - one syncing small videos (15000 ~100mb files), one syncing photos (15000 ~5mb files) and another one syncing a 3GB volume with about 100.000 small (10-500kb) files. The high io wait situation occurred after about 3 hours of processing. I use newrelic to monitor the server, so I could notice a spike of writes on my local cache disk, which I documented here - https://docs.google.com/document/d/1S927JPyMG4SCxkiIcReQDHWACM1qozYOMTa86pGQ1k4/edit?usp=sharing What does ~/.s3ql/mount.log say? Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Reduced file handles?
Rabid Mutant rabidmutantstarg...@gmail.com writes: On Saturday, January 24, 2015 at 3:00:20 AM UTC+11, Nikolaus Rath wrote: On 01/21/2015 08:16 PM, Rabid Mutant wrote: Does S3QL really need 1 file handle per cache entry? In principle, no. The way it's currently programmed, yes. Looking briefly at the code, it seems I might be able to replace access to the file handle with a call to a cache manager, and everything should just work...but that's based on one quick look. Would that be a correct assessment? It sounds right, but I haven't looked in detail either. I could use rsync to compare files and update only the new changed files without any unnecessary network I/O. It would also allow for the possibility of offline use. rsync by default uses file name, modification time, and size to check if a file has changed, so it won't incur any network IO apart from what's necessary to transfer new and changed files. This changes if you use the -c option, but I'd be rather curious why you'd need that. Some applications (notably PostgreSQL) do not update inode dates when they update files, specifically to reduce IO load. ie. the data is changed, but the modification dates (and quite probably size) are not. Are you sure that's correct? Updating the inode dates happens in the kernel, and as far as I know there is no way for an userspace application to prevent this. You can of course reset the times to the original values, but this would increase the IO load rather than reduce it. So the -c option becomes important, at least in this case. I also (sometimes) change the file modification dates on photos to the original photo date after trivial edits: eg changing EXIF data. In this case the date and size remain the same. That might cause problems indeed. But seriously, I'd simply not do that instead of trying to patch S3QL to support a bigger cache. Another factor, and I agree it's probably minor, but decryption is usually considerably faster than compression, and my expectation was that using 'rsync -c' on a fully cached file system (thereby comparing uncompressed data) would be faster than compressing the data and comparing the checksums. The checksum is calculated before compression. Compression and encryption only happens after the checksum has been calculated and no matching existing block has been found. Since the following would occur: Normal Copy: 1. compress chunk 2. compare hash in DB 3. If different: a. send compressed from step 1 No. It's 1. Calculate checksum 2. If different: a. compress b. encrypt c. upload rsync -c copy with complete cache: 1. decompress chunk from cache 2. compare data 3. if different: a. Compress chunk b. send No, that wrong as well. The cache is stored uncompressed. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Uncaught top-level exception - ver 1.17 on Ubuntu 12.04LTS
Hi Warren, Warren Daly invisibleagentreco...@gmail.com writes: [...] File /usr/local/lib/python2.7/dist-packages/s3ql-1.17-py2.7-linux-x86_64.egg /s3ql/backends/s3c.py, line 281, in open_read resp = self._do_request('GET', '/%s%s' % (self.prefix, key)) File /usr/local/lib/python2.7/dist-packages/s3ql-1.17-py2.7-linux-x86_64.egg /s3ql/backends/s3c.py, line 405, in _do_request tree = ElementTree.parse(resp).getroot() File string, line 62, in parse File string, line 38, in parse ParseError: no element found: line 1, column 0 This looks as if Amazon send you a malformed XML error message. If you use a current S3QL version, it'd give more debugging information so that we could find out if it's truly an Amazon S3 problem (S3QL development has uncovered quite a few of those so far) and if there might be a workaround. As long as you stay with 1.x, though, the chances of doing anything about this are pretty slim. I don't run that version on any of my systems anymore, the required changes are probably non-trivial, and this seems like a pretty rare occurence, so I'm unlikely to spend much time on it. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Corrupt metadata
Hi Eric, When replying to emails on this list, please do not put your reply above the quoted text, and do not quote the entire message you're answering to. This makes it unnecessarily hard for other readers to understand the context of your email. Instead, please cut quoted parts that are not relevant to your reply, and insert your responses right after the points you're replying to (as I have done below). Thanks! Eric Eijkelenboom eric.eijkelenb...@gmail.com writes: Hi Nikolaus Thanks a lot for your advice. I will definitely refrain from messing with the metadata in the future :) I renamed s3ql_metadata to something not starting with 's3ql', but running: fsck.s3ql s3://mybucket results in this exception: Starting fsck of s3://mybucket Uncaught top-level exception: Traceback (most recent call last): File /usr/lib/s3ql/s3ql/backends/s3c.py, line 253, in lookup resp = self._do_request('HEAD', '/%s%s' % (self.prefix, key)) File /usr/lib/s3ql/s3ql/backends/s3c.py, line 477, in _do_request self._parse_error_response(resp) File /usr/lib/s3ql/s3ql/backends/s3c.py, line 494, in _ parse_error_response raise HTTPError(resp.status, resp.reason, resp.headers) s3ql.backends.s3c.HTTPError: 404 Not Found Aeh, yeah, I forgot about that. fsck.s3ql doesn't cope well with completely absent metadata either... Can you temporarily apply the attached patch? Hopefully that will fix your problem. Just make sure to use it only once, and then revert to the vanilla 2.12 version. Best, Nikolaus PS: Please remember the first paragraph when replying :-). -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout. diff --git a/src/s3ql/fsck.py b/src/s3ql/fsck.py --- a/src/s3ql/fsck.py +++ b/src/s3ql/fsck.py @@ -1159,7 +1159,7 @@ param = pickle.load(fh) if param['seq_no'] seq_no: log.info('Ignoring locally cached metadata (outdated).') -param = backend.lookup('s3ql_metadata') +param = backend.lookup('s3ql_metadata_bak_0') else: log.info('Using cached metadata.') db = Connection(cachepath + '.db') @@ -1169,12 +1169,12 @@ log.warning('File system has not been unmounted cleanly.') param['needs_fsck'] = True -elif backend.lookup('s3ql_metadata')['seq_no'] != param['seq_no']: +elif backend.lookup('s3ql_metadata_bak_0')['seq_no'] != param['seq_no']: log.warning('Remote metadata is outdated.') param['needs_fsck'] = True else: -param = backend.lookup('s3ql_metadata') +param = backend.lookup('s3ql_metadata_bak_0') assert not os.path.exists(cachepath + '-cache') # .db might exist if mount.s3ql is killed at exactly the right instant # and should just be ignored. @@ -1287,6 +1287,7 @@ is_compressed=True) log.info('Wrote %s of compressed metadata.', pretty_print_size(obj_fh.get_obj_size())) log.info('Cycling metadata backups...') +backend.copy('s3ql_metadata_new', 's3ql_metadata') cycle_metadata(backend) with open(cachepath + '.params', 'wb') as fh: pickle.dump(param, fh, PICKLE_PROTOCOL)
Re: [s3ql] Reduced file handles?
Rabid Mutant rabidmutantstarg...@gmail.com writes: I must admit I see the 'rsync -c' option as a much stronger guarantee than a copy based on file dates sizes, and this could be irrational. But, if 100% cache can be achieved at nearly zero performance impact *and* it supports rsync -c as well as an offline mode, this seems to me like a big gain at minimal cost. I think what you want in this case is use something like lsyncd. It uses inotify to determine changed files. So it can't miss anything even if you mess with file time stamps, and it does not require you to read the entire file: See also: https://bitbucket.org/nikratio/s3ql/wiki/FAQ#!is-there-a-way-to-make-the-cache-persistent-access-the-file-system-offline Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Corrupt metadata
Eric Eijkelenboom eric.eijkelenb...@gmail.com writes: Hi guys Here’s what happened: 1. Due to a server shutdown, the filesystem was not cleanly umount’ed. Hereafter, fsck resulted in ‘fs still mounted elsewhere, aborting’ errors (obviously, the backend was not mounted anywhere else). Did you run with --batch? Normally, fsck will prompt you and you can tell it to continue running anyway. No need to mess with the metadata at all. 2. I tried restoring from backup by renaming s3://mybucket/s3ql_metadata_bak_0 to to s3://mybucket/s3ql_metadata, hereby overwriting s3ql_metadata (yes, big screw-up). Yep. That breaks things, because it looks as if some malicious other guy is trying to get you to use outdated metadata. After discovering s3qladm download-metadata, I tried to: 1. s3qladm download-metadata s3://mybucket 2. Choose a backup 3. Move the generated files to ~/.s3ql 4. Run fsck. fsck does it’s thing, but crashes in the end with the same exception as above: ... Backing up old metadata... Uncaught top-level exception: Traceback (most recent call last): File /usr/bin/fsck.s3ql, line 9, in module load_entry_point('s3ql==2.12', 'console_scripts', 'fsck.s3ql')() File /usr/lib/s3ql/s3ql/fsck.py, line 1291, in main cycle_metadata(backend) File /usr/lib/s3ql/s3ql/metadata.py, line 127, in cycle_metadata backend.copy(s3ql_metadata, s3ql_metadata_bak_0) File /usr/lib/s3ql/s3ql/backends/comprenc.py, line 290, in copy self._copy_or_rename(src, dest, rename=False, metadata=metadata) File /usr/lib/s3ql/s3ql/backends/comprenc.py, line 302, in _ copy_or_rename (nonce, meta_old) = self._verify_meta(src, meta_raw) File /usr/lib/s3ql/s3ql/backends/comprenc.py, line 139, in _ verify_meta % (stored_key, key)) s3ql.backends.common.CorruptedObjectError: Object content does not match its key (s3ql_metadata_new vs s3ql_metadata) Questions: 1. What can I do from here? Delete the s3ql_metadata object from the bucket (or rename it to something that does not start with s3ql_). 2. How to avoid this in the future? * Never, ever, modify the bucket contents with anything but the S3QL tools. * Don't use s3qladm download-metadata before you've asked on this list, fsck.s3ql should be able to handle 99% of all problems. * Don't shut down the server while the file system is still mounted :-) Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Trying to use s3ql to attach to Swift
On Mar 31 2015, Shane Wilson coswort...@gmail.com wrote: First, I'm very new to Swift, so I'm sure I'm doing something wrong. I'm trying to mount swift to Ubuntu 14.04 linux using s3ql, but I getting an error during the mkfs.s3ql. Here is the command line: root@owncloud:~/.s3ql# mkfs.s3ql --plain --backend-options no-ssl swiftks://192.102.218.230:5000/regionOne:s3ql Enter backend login: Enter backend passphrase: No accessible object storage service found in region regionOne (available regions: ) root@owncloud:~/.s3ql# Here is my authinfo2 [swift] backend-login: tenant:demo backend-password: 19monza67 storage-url: swiftks://192.102.218.230:5000/regionOne:s3ql The s3ql container is there when I do a swift list. Its as if it doesn't know about the region. There should only be one region, my swift install is from the Juno install docs and only creates regionOne. I hope you can point me in the right direction, I'm sure it something simple I'm missing. I've read everything and I'm at a lose for direction. Apply the attached patch and post the contents of the response.txt file that it should create. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout. diff --git a/src/s3ql/adm.py b/src/s3ql/adm.py --- a/src/s3ql/adm.py +++ b/src/s3ql/adm.py @@ -398,6 +398,8 @@ def yield_objects(): for (id_,) in db.query('SELECT id FROM objects'): +if id_ my_starting_id: +continue yield 's3ql_data_%d' % id_ for obj_id in extra_objects: yield obj_id diff --git a/src/s3ql/backends/swiftks.py b/src/s3ql/backends/swiftks.py --- a/src/s3ql/backends/swiftks.py +++ b/src/s3ql/backends/swiftks.py @@ -97,7 +97,10 @@ elif resp.status 299 or resp.status 200: raise HTTPError(resp.status, resp.reason, resp.headers) -cat = json.loads(conn.read().decode('utf-8')) +body = conn.read() +with open('response.txt', 'wb') as fh: +fh.write(body) +cat = json.loads(body.decode('utf-8')) self.auth_token = cat['access']['token']['id'] avail_regions = []
Re: [s3ql] Re: Preventing huge network traffic
[Quoting repaired] On Apr 20 2015, Viktor Szépe szepe.vik...@gmail.com wrote: On Saturday, April 18, 2015 at 10:53:05 AM UTC+2, Viktor Szépe wrote: Is there a way to upload only changed file, thus doing incremental backup? My usecase is making backup of a website that has 1 or two new files per backup. Excuse me that was a wrong question. The right one is: Will s3ql read the file's contents to determine whether it has change or reads only the file's metadata (mtime, size etc.)? I don't think that's the right question either. S3QL is a file system. When a client application (like your backup program) tells it to write data to a file, it checks if it already has identical data stored and (if so) saves only one copy. There is no way for S3QL itself to compare mtime and size. S3QL does not know when a file has reached its final size or mtime, and the mtime is the time of last write, i.e. when S3QL receives a write request it is always the current time. Finally, S3QL also has no idea which other file to compare any of this against. What you want is a feature of the backup program. For example, if you use rsync it will only copy a file if its attributes have changed. Look at http://www.rath.org/s3ql-docs/contrib.html#s3-backup-sh for an example. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Storage backend hung
On Apr 28 2015, netplus.r...@gmail.com wrote: Hmm. After a 502 error S3QL tries to re-establish the connection Completely. The only difference is that it does not attempt to contact The auth server again. Could it be that the auth server sends you to a Different endpoint when you restart S3QL? If you run a traffic dump tool like Wireshark, you should be able to see If S3QL connects to the same server when you restart. Yes, there is indeed two VIP for the auth part and the storage part. Is there any workaround to force the connection to contact the auth server ? No, but it might be worth implementing. Could you file an issue at https://bitbucket.org/nikratio/s3ql/issues? Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] S3 performance on lots small files
On Apr 28 2015, Maxim Kostrikin mkostri...@gmail.com wrote: Hello, I am faced with performance issue. My s3ql mount attached to s3 backend with options: --cachesize 2000 --max-cache-entries 100 --threads 16 --allow-other --backend-options no-ssl,tcp-timeout=30 --nfs I copy into s3ql mount many small files 100 - 1000 bytes. With time copy speed slowdown almost stall. Please be more precise. How did you measure the speed? How fast was it initially, how fast at the end? I see almost 16 connections with s3 servers, but tcpdump shows small rate of uploading PUTs. What do you mean with small? What do you expect it to be, and what did you find instead? Generally every file needs one network request, which means tens to hundreds of milliseconds of network latency per file. Ok, I remounted with --threads 1024 ( I bet it was a bad idea), but connections with s3 was 24 max and lots of cross locks between threads. How did you determine that? 1. How to increase parallels request to s3? Or increase upload performance? In theory more threads should give you more simulateous connection - unless you're limited by something else. 2. What limits could be reasonable for --threads parameter? Why? SQLite bottleneck? Difficult to say without further information. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] S3 performance on lots small files
On May 04 2015, Maxim Kostrikin mkostri...@gmail.com wrote: I see almost 16 connections with s3 servers, but tcpdump shows small rate of uploading PUTs. What do you mean with small? What do you expect it to be, and what did you find instead? 200 bytes - 1k per file. The dirty size was big (1-2GiB). No, I was talking about the small rate of PUTs that you saw in tcpdump. If we assume about 100 ms network latency, then with 200 byte files you'd be able to transfer at most 200 bytes / 100 ms = 2 kB/s for each active thread. So your 5 kb/s isn't completely the wrong order of magnitude. I would expect that the initial fast period is just when you file up your cache. Am I right that, if you set the cache to something small (say 2 * number of threads entries), you're getting a constant transfer rate from the very beginning? In tcpdump, how many PUTs do you see per second? Is that number constant? All right, but why only 16 threads were transferring? I could try to determine that, if you'd answer my question... Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] S3 performance on lots small files
On Apr 30 2015, Maxim Kostrikin mkostri...@gmail.com wrote: среда, 29 апреля 2015 г., 2:55:05 UTC+6 пользователь Nikolaus Rath написал: On Apr 28 2015, Maxim Kostrikin mkost...@gmail.com javascript: wrote: I am faced with performance issue. My s3ql mount attached to s3 backend with options: --cachesize 2000 --max-cache-entries 100 --threads 16 --allow-other --backend-options no-ssl,tcp-timeout=30 --nfs I copy into s3ql mount many small files 100 - 1000 bytes. With time copy speed slowdown almost stall. Please be more precise. How did you measure the speed? How fast was it initially, how fast at the end? I was transfering files with tar c|dd|ssh s3qluser@s3ql_host tar xC /s3ql_mount kill -USR1 $dd_pid shows the speed. It starts from 5M/sec and falls to 5k/s or even freeze. Well, did it fall to 5 k/s or did it freeze? Now I tar-pipe into temp directory and from the temp directory move into /s3ql_mount I see almost 16 connections with s3 servers, but tcpdump shows small rate of uploading PUTs. What do you mean with small? What do you expect it to be, and what did you find instead? 200 bytes - 1k per file. The dirty size was big (1-2GiB). No, I was talking about the small rate of PUTs that you saw in tcpdump. If we assume about 100 ms network latency, then with 200 byte files you'd be able to transfer at most 200 bytes / 100 ms = 2 kB/s for each active thread. So your 5 kb/s isn't completely the wrong order of magnitude. I would expect that the initial fast period is just when you file up your cache. Am I right that, if you set the cache to something small (say 2 * number of threads entries), you're getting a constant transfer rate from the very beginning? In tcpdump, how many PUTs do you see per second? Is that number constant? Ok, I remounted with --threads 1024 ( I bet it was a bad idea), but connections with s3 was 24 max and lots of cross locks between threads. How did you determine that? strace -fp $mount.s3ql_pid showed many FUTEX messages. I think that doesn't mean that there's lock contention, just that locks are being used. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Not detecting mounted elsewhere
On May 08 2015, Andy Cress andycr...@gmail.com wrote: I’m using s3ql-1.17, and this pertains to Amazon S3. I configured an S3QL mount point with mkfs.s3ql and mount.s3ql on one system, then I did fsck.s3ql and mount.s3ql to the same bucket (same credentials) on a second system, but it worked and did not give me the ‘still mounted elsewhere’ message that I expected. What functions should I use instead to detect this case? Or should this work in fsck.s3ql / mount.s3ql in a later version? What region is your bucket in? S3QL tries to detect double mounts as best as it can, but there are fundamental limits imposed by the storage backend. For the detection to work reliably, the backend must guarantee either immediate get-after-put or list-after-put consistency. The us-standard storage region, for example, offers neither. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] Understanding benchmark.py output
On May 13 2015, joshua.phill...@semanticbits.com wrote: Hi, I'm trying to understand the benchmark.py output to identify an upload bottleneck in my S3QL deployment. I'd appreciate any pointers you may have. [...] Threads: 1 2 4 8 Max FS throughput (bzip2): 36264 KiB/s 36264 KiB/s 36264 KiB/s 36264 ..limited by: S3QL/FUSE S3QL/FUSE S3QL/FUSE S3QL/FUSE But, when I run when I run the mount with 1MB cache size, 8 upload thread, and using bzip2-6 compression, I'm seeing a much lower throughput than expected. mount.s3ql --threads 8 --nfs --allow-other --cachedir /var/lib/s3ql-cache --cachesize 1024 --compress bzip2-6 --backend-options no-ssl s3://some-bucket /some/path Autodetected 4040 file descriptors available for cache entries Using cached metadata. Creating NFS indices... Mounting filesystem... dd if=/dev/zero of=/some/path/speed_test.dat bs=2M count=1 1+0 records in 1+0 records out 2097152 bytes (2.1 MB) copied, 33.6993 s, 62.2 kB/s Only, 62 KB/s. Any ideas why it's so low, or where to look for the bottleneck? That's odd. What happens if you increase the cache size to 4 MB? What happens if you decrease the blocksize (bs) to 128k? Also, when I try the same thing from my home laptop, I see throughput of 150 MB/s, which is my upload limit, as expected. I don't believe it. Please recheck. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] load balancing/fail over for NFS with s3ql on swift
On May 15 2015, Jeb Baxley jeb.bax...@gmail.com wrote: So I'm looking at using S3QL on swift, and was wondering if it's possible to provide a NFS cluster using s3ql on swift? Depends what you mean by NFS cluster. You can export an S3QL mountpoint over NFS. Thinking about it, I'm not sure it would be given how both of these technologies work - but was wondering if it's possible to provide a customer/client with an nfs target that is serviced by more than one s3ql mount/server instance. I've never heard of having one NFS file system served by multiple servers. Are you sure that's possible? Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [s3ql] max-obj-size with OpenStack / Swift
On May 13 2015, Maxim Kostrikin mkostri...@gmail.com wrote: I think the only reason for bigger max-obj-size is a db size. But 62.8 MiB is considerably small. I have a setup with 10Gb+ of sqlite DB and it works, I guess the cons of bigger max-obj-size is multicore processing. If block is big, then s3ql should wait till next big block will be copied to start using next core. Or not? That's correct. Best, -Nikolaus -- GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F »Time flies like an arrow, fruit flies like a Banana.« -- You received this message because you are subscribed to the Google Groups s3ql group. To unsubscribe from this group and stop receiving emails from it, send an email to s3ql+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.