Re: svn commit: r247485 - in stable/9: crypto/openssh crypto/openssh/openbsd-compat secure/lib/libssh secure/usr.sbin/sshd
On 3/1/2013 10:06 PM, Mike Tancsa wrote: On 3/1/2013 3:34 PM, Dag-Erling Smørgrav wrote: Mike Tancsa m...@sentex.net writes: Dag-Erling Smørgrav d...@des.no writes: Are you sure this was due to the OpenSSH update, and not the OpenSSL update a few days ago? Can you try to roll back to r247484? I didnt think openssl got updated on RELENG_9 ? Ah, you're right. There is an OpenSSL commit immediately before my OpenSSH commit in src/secure, but it's from last July :) Can you try to connect against each version in turn while running tcpdump or wireshark and show me the pre-kex handshake and proposal exchange (basically, everything that's transmitted in cleartext) in both cases? The pcaps and basic wireshark output at http://tancsa.com/openssh/ This PR looks to be related http://lists.freebsd.org/pipermail/freebsd-bugs/2012-September/050139.html ---Mike -- --- Mike Tancsa, tel +1 519 651 3400 Sentex Communications, m...@sentex.net Providing Internet services since 1994 www.sentex.net Cambridge, Ontario Canada http://www.tancsa.com/ ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to freebsd-stable-unsubscr...@freebsd.org
Re: Musings on ZFS Backup strategies
On Fri, 01 Mar 2013 21:34:39 +0100, Daniel Eischen deisc...@freebsd.org wrote: On Fri, 1 Mar 2013, Ben Morrow wrote: Quoth Daniel Eischen deisc...@freebsd.org: Yes, we still use a couple of DLT autoloaders and have nightly incrementals and weekly fulls. This is the problem I have with converting to ZFS. Our typical recovery is when a user says they need a directory or set of files from a week or two ago. Using dump from tape, I can easily extract *just* the necessary files. I don't need a second system to restore to, so that I can then extract the file. As Karl said originally, you can do that with snapshots without having to go to your backups at all. With the right arrangements (symlinks to the .zfs/snapshot/* directories, or just setting the snapdir property to 'visible') you can make it so users can do this sort of restore themselves without having to go through you. It wasn't clear that snapshots were traversable as a normal directory structure. I was thinking it was just a blob that you had to roll back to in order to get anything out of it. That is the main benefit of snapshots. :-) You can also very easily diff files between them. Mostly a lot of data is static so it does not cost a lot to keep snapshots. There are a lot of scripts online and in ports which make a nice retention policy like e.g. 7 daily snaphots, 8 weekly, 12 monthly, 2 yearly. See below for (an incomplete list of) what I keep about my homedir at home. Under our current scheme, we would remove snapshots after the next (weekly) full zfs send (nee dump), so it wouldn't help unless we kept snapshots around a lot longer. Why not. Am I correct in assuming that one could: # zfs send -R snapshot | dd obs=10240 of=/dev/rst0 to archive it to tape instead of another [system:]drive? Yes, your are correct. The manual page about zfs send says: 'The format of the stream is committed. You will be able to receive your streams on future versions of ZFS.' Ronald. tank/home 115G 65.6G 53.6G /home tank/home@auto-2011-10-25_19.00.yearly 16.3G - 56.8G - tank/home@auto-2012-06-06_22.00.yearly 5.55G - 53.3G - tank/home@auto-2012-09-02_20.00.monthly 2.61G - 49.3G - tank/home@auto-2012-10-15_06.00.monthly 2.22G - 49.9G - tank/home@auto-2012-11-26_13.00.monthly 2.47G - 50.2G - tank/home@auto-2013-01-07_13.00.monthly 2.56G - 51.5G - tank/home@auto-2013-01-21_13.00.weekly 1.06G - 52.4G - tank/home@auto-2013-01-28_13.00.weekly409M - 52.3G - tank/home@auto-2013-02-04_13.00.monthly 625M - 52.5G - tank/home@auto-2013-02-11_13.00.weekly689M - 52.5G - tank/home@auto-2013-02-16_13.00.weekly 17.7M - 52.5G - tank/home@auto-2013-02-17_13.00.daily17.7M - 52.5G - tank/home@auto-2013-02-18_13.00.daily17.9M - 52.5G - ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to freebsd-stable-unsubscr...@freebsd.org
Make use of RACCT and rctl
Hi! Im trying to limit memory usage for jails with the rctl API. But I don't really get it. I have compiled the kernel with the right options and rctl show me stuff like: jail:jail22:memoryuse:deny=268435456 jail:jail22:swapuse:deny=268435456 jail:jail20:memoryuse:deny=268435456 jail:jail20:swapuse:deny=268435456 jail:jail16:memoryuse:deny=268435456 jail:jail16:swapuse:deny=268435456 but when I try to allocate memory it doesn't seem to hit the limit. Also when I run # rctl -u jail:jail20 cputime=0 datasize=0 stacksize=0 coredumpsize=0 memoryuse=0 memorylocked=0 maxproc=0 openfiles=0 vmemoryuse=0 pseudoterminals=0 swapuse=0 nthr=0 msgqqueued=0 msgqsize=0 nmsgq=0 nsem=0 nsemop=0 nshm=0 shmsize=0 wallclock=0 it seems that no accounting is done. What's missing? Cant find anything in the manuals. # uname -srm FreeBSD 9.1-RELEASE-p1 amd64 /Peter. ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to freebsd-stable-unsubscr...@freebsd.org
Re: Musings on ZFS Backup strategies
On Fri, 01 Mar 2013 18:55:22 +0100, Volodymyr Kostyrko c.kw...@gmail.com wrote: 01.03.2013 16:24, Karl Denninger: Dabbling with ZFS now, and giving some thought to how to handle backup strategies. ZFS' snapshot capabilities have forced me to re-think the way that I've handled this. Previously near-line (and offline) backup was focused on being able to handle both disasters (e.g. RAID adapter goes nuts and scribbles on the entire contents of the array), a double-disk (or worse) failure, or the obvious (e.g. fire, etc) along with the aw crap, I just rm -rf'd something I'd rather not! ZFS makes snapshots very cheap, which means you can resolve the aw crap situation without resorting to backups at all. This turns the backup situation into a disaster recovery one. And that in turn seems to say that the ideal strategy looks more like: Take a base snapshot immediately and zfs send it to offline storage. Take an incremental at some interval (appropriate for disaster recovery) and zfs send THAT to stable storage. If I then restore the base and snapshot, I get back to where I was when the latest snapshot was taken. I don't need to keep the incremental snapshot for longer than it takes to zfs send it, so I can do: zfs snapshot pool/some-filesystem@unique-label zfs send -i pool/some-filesystem@base pool/some-filesystem@unique-label zfs destroy pool/some-filesystem@unique-label and that seems to work (and restore) just fine. Yes, I'm working with backups the same way, I wrote a simple script that synchronizes two filesystems between distant servers. I also use the same script to synchronize bushy filesystems (with hundred thousands of Your filesystems grow a lot of hair? :-) files) where rsync produces a too big load for synchronizing. https://github.com/kworr/zfSnap/commit/08d8b499dbc2527a652cddbc601c7ee8c0c23301 I left it where it was but I was also planning to write some purger for snapshots that would automatically purge snapshots when pool gets low on space. Never hit that yet. Am I looking at this the right way here? Provided that the base backup and incremental are both readable, it appears that I have the disaster case covered, and the online snapshot increments and retention are easily adjusted and cover the oops situations without having to resort to the backups at all. This in turn means that keeping more than two incremental dumps offline has little or no value; the second merely being taken to insure that there is always at least one that has been written to completion without error to apply on top of the base. That in turn makes the backup storage requirement based only on entropy in the filesystem and not time (where the tower of Hanoi style dump hierarchy imposed both a time AND entropy cost on backup media.) Well, snapshots can pose a value in a longer timeframe depending on data. Being able to restore some file accidentally deleted two month ago already saved 2k$ for one of our customers. ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to freebsd-stable-unsubscr...@freebsd.org
Re: svn commit: r247485 - in stable/9: crypto/openssh crypto/openssh/openbsd-compat secure/lib/libssh secure/usr.sbin/sshd
Mike Tancsa m...@sentex.net writes: This PR looks to be related http://lists.freebsd.org/pipermail/freebsd-bugs/2012-September/050139.html That suggests a bug in the aesni driver... Can you ktrace sshd in both cases? My guess is the difference is that the new version uses hw offloading while the old version doesn't. DES -- Dag-Erling Smørgrav - d...@des.no ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to freebsd-stable-unsubscr...@freebsd.org
Re: svn commit: r247485 - in stable/9: crypto/openssh crypto/openssh/openbsd-compat secure/lib/libssh secure/usr.sbin/sshd
On 3/2/2013 10:33 AM, Dag-Erling Smørgrav wrote: Mike Tancsa m...@sentex.net writes: This PR looks to be related http://lists.freebsd.org/pipermail/freebsd-bugs/2012-September/050139.html That suggests a bug in the aesni driver... OK, but the above uses the glxsb driver, not the aesni driver. Also, if at was aesni, would it not show up in the geli tests I did ? Can you ktrace sshd in both cases? My guess is the difference is that the new version uses hw offloading while the old version doesn't. Done. Both files are at http://www.tancsa.com/openssh ---Mike -- --- Mike Tancsa, tel +1 519 651 3400 Sentex Communications, m...@sentex.net Providing Internet services since 1994 www.sentex.net Cambridge, Ontario Canada http://www.tancsa.com/ ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to freebsd-stable-unsubscr...@freebsd.org
Re: svn commit: r247485 - in stable/9: crypto/openssh crypto/openssh/openbsd-compat secure/lib/libssh secure/usr.sbin/sshd
Mike Tancsa m...@sentex.net writes: The pcaps and basic wireshark output at http://tancsa.com/openssh/ This is 6.1 with aesni vs 6.1 without aesni; what I wanted was 6.1 vs 5.8, both with aesni loaded. Could you also ktrace the server in both cases? An easy workaround is to change the list of ciphers the server will offer to clients by adding a Ciphers line in /etc/ssh/sshd_config. The default is: Ciphers aes128-ctr,aes192-ctr,aes256-ctr,arcfour256,arcfour128,aes128-cbc,3des-cbc,blowfish-cbc,cast128-cbc,aes192-cbc,aes256-cbc,arcfour Either remove the AES entries or move them further down the list. The client will normally pick the first supported cipher. As far as I can tell, SecureCRT supports all the same ciphers that OpenSSH does, so just moving arcfour{256,128} to the front of the list should work. (AFAIK, arcfour is also much faster than aes) DES -- Dag-Erling Smørgrav - d...@des.no ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to freebsd-stable-unsubscr...@freebsd.org
Re: svn commit: r247485 - in stable/9: crypto/openssh crypto/openssh/openbsd-compat secure/lib/libssh secure/usr.sbin/sshd
On 3/2/2013 11:02 AM, Dag-Erling Smørgrav wrote: Mike Tancsa m...@sentex.net writes: The pcaps and basic wireshark output at http://tancsa.com/openssh/ This is 6.1 with aesni vs 6.1 without aesni; what I wanted was 6.1 vs 5.8, both with aesni loaded. Ahh, ok. I will do it later this aft. Could you also ktrace the server in both cases? That was the daemon in both cases. ktrace /usr/sbin/sshd - An easy workaround is to change the list of ciphers the server will offer to clients by adding a Ciphers line in /etc/ssh/sshd_config. The default is: Ciphers aes128-ctr,aes192-ctr,aes256-ctr,arcfour256,arcfour128,aes128-cbc,3des-cbc,blowfish-cbc,cast128-cbc,aes192-cbc,aes256-cbc,arcfour Either remove the AES entries or move them further down the list. The client will normally pick the first supported cipher. As far as I can tell, SecureCRT supports all the same ciphers that OpenSSH does, so just moving arcfour{256,128} to the front of the list should work. (AFAIK, arcfour is also much faster than aes) Actually, I am just doing with a freebsd openssh client ssh -c aes128-cbc testhost-with-the-issue.sentex.ca Its for sure something to do with hardware crypto offload because it works fine with a cipher that is not accelerated. ---Mike DES -- --- Mike Tancsa, tel +1 519 651 3400 Sentex Communications, m...@sentex.net Providing Internet services since 1994 www.sentex.net Cambridge, Ontario Canada http://www.tancsa.com/ ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to freebsd-stable-unsubscr...@freebsd.org
Re: svn commit: r247485 - in stable/9: crypto/openssh crypto/openssh/openbsd-compat secure/lib/libssh secure/usr.sbin/sshd
Dag-Erling Smørgrav d...@des.no writes: This is 6.1 with aesni vs 6.1 without aesni; what I wanted was 6.1 vs 5.8, both with aesni loaded. On second thought, I don't need more pcaps. DES -- Dag-Erling Smørgrav - d...@des.no ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to freebsd-stable-unsubscr...@freebsd.org
Re: Make use of RACCT and rctl
Wiadomość napisana przez Peter Ankerstål w dniu 2 mar 2013, o godz. 16:21: Hi! Im trying to limit memory usage for jails with the rctl API. But I don't really get it. I have compiled the kernel with the right options and rctl show me stuff like: jail:jail22:memoryuse:deny=268435456 jail:jail22:swapuse:deny=268435456 jail:jail20:memoryuse:deny=268435456 jail:jail20:swapuse:deny=268435456 jail:jail16:memoryuse:deny=268435456 jail:jail16:swapuse:deny=268435456 but when I try to allocate memory it doesn't seem to hit the limit. Also when I run # rctl -u jail:jail20 cputime=0 [..] Could you please do jls jid name and verify that a jail named jail20 is actually running? -- If you cut off my head, what would I say? Me and my head, or me and my body? ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to freebsd-stable-unsubscr...@freebsd.org
Re: Make use of RACCT and rctl
On Mar 2, 2013, at 5:15 PM, Edward Tomasz Napierała tr...@freebsd.org wrote: [..] Could you please do jls jid name and verify that a jail named jail20 is actually running? -- If you cut off my head, what would I say? Me and my head, or me and my body? Oh! My bad, I thought it was the name from rc.conf. But of course it is the name from the -n flag. Now everything seems to work. Thanks! ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to freebsd-stable-unsubscr...@freebsd.org
Re: svn commit: r247485 - in stable/9: crypto/openssh crypto/openssh/openbsd-compat secure/lib/libssh secure/usr.sbin/sshd
On Sat, 2013-03-02 at 17:02 +0100, Dag-Erling Smørgrav wrote: Mike Tancsa m...@sentex.net writes: The pcaps and basic wireshark output at http://tancsa.com/openssh/ This is 6.1 with aesni vs 6.1 without aesni; what I wanted was 6.1 vs 5.8, both with aesni loaded. Could you also ktrace the server in both cases? An easy workaround is to change the list of ciphers the server will offer to clients by adding a Ciphers line in /etc/ssh/sshd_config. The default is: Ciphers aes128-ctr,aes192-ctr,aes256-ctr,arcfour256,arcfour128,aes128-cbc,3des-cbc,blowfish-cbc,cast128-cbc,aes192-cbc,aes256-cbc,arcfour Either remove the AES entries or move them further down the list. The client will normally pick the first supported cipher. As far as I can tell, SecureCRT supports all the same ciphers that OpenSSH does, so just moving arcfour{256,128} to the front of the list should work. (AFAIK, arcfour is also much faster than aes) The last time I tried to affect the chosen cypher by manipulating the order of the list items in the config files was a couple years ago, but I found then that you just can't do that. The client side, not the server, decides on the order, and it's based on compiled-in ordering within the client code (not the client config). From the server side the only thing you can do to affect the order is leave items out of the list (it will still try the remaining list items in the client-requested order). All of this was with OpenSSH_5.4p1_hpn13v11 FreeBSD-20100308, OpenSSL 0.9.8q 2 Dec 2010 and may be completely out of date now. -- Ian ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to freebsd-stable-unsubscr...@freebsd.org
Re: Make use of RACCT and rctl
Wiadomość napisana przez Peter Ankerstål w dniu 2 mar 2013, o godz. 17:18: On Mar 2, 2013, at 5:15 PM, Edward Tomasz Napierała tr...@freebsd.org wrote: [..] Could you please do jls jid name and verify that a jail named jail20 is actually running? -- If you cut off my head, what would I say? Me and my head, or me and my body? Oh! My bad, I thought it was the name from rc.conf. But of course it is the name from the -n flag. Now everything seems to work. Thanks! You're welcome. Please Cc: me if you have any further questions or comments. -- If you cut off my head, what would I say? Me and my head, or me and my body? ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to freebsd-stable-unsubscr...@freebsd.org
Re: Musings on ZFS Backup strategies
On Mar 1, 2013, at 21:14, Ben Morrow wrote: But since ZFS doesn't support POSIX.1e ACLs that's not terribly useful... I don't believe bsdtar/libarchive supports NFSv4 ACLs yet. Ah yes, just noticed that. Thought it did. https://github.com/libarchive/libarchive/wiki/TarNFS4ACLs ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to freebsd-stable-unsubscr...@freebsd.org
Re: Musings on ZFS Backup strategies
On 2013-Mar-01 08:24:53 -0600, Karl Denninger k...@denninger.net wrote: If I then restore the base and snapshot, I get back to where I was when the latest snapshot was taken. I don't need to keep the incremental snapshot for longer than it takes to zfs send it, so I can do: zfs snapshot pool/some-filesystem@unique-label zfs send -i pool/some-filesystem@base pool/some-filesystem@unique-label zfs destroy pool/some-filesystem@unique-label and that seems to work (and restore) just fine. This gives you an incremental since the base snapshot - which will probably grow in size over time. If you are storing the ZFS send streams on (eg) tape, rather than receiving them, you probably still want the Towers of Hanoi style backup hierarchy to control your backup volume. It's also worth noting that whilst the stream will contain the compression attributes of the filesystem(s) in it, the actual data is the stream in uncompressed This in turn means that keeping more than two incremental dumps offline has little or no value; the second merely being taken to insure that there is always at least one that has been written to completion without error to apply on top of the base. This is quite a critical point with this style of backup: The ZFS send stream is not intended as an archive format. It includes error detection but no error correction and any error in a stream renders the whole stream unusable (you can't retrieve only part of a stream). If you go this way, you probably want to wrap the stream in a FEC container (eg based on ports/comms/libfec) and/or keep multiple copies. The recommended approach is to do zfs send | zfs recv and store a replica of your pool (with whatever level of RAID that meets your needs). This way, you immediately detect an error in the send stream and can repeat the send. You then use scrub to verify (and recover) the replica. (Yes, I know, I've been a ZFS resister ;-)) Resistance is futile. :-) On 2013-Mar-01 15:34:39 -0500, Daniel Eischen deisc...@freebsd.org wrote: It wasn't clear that snapshots were traversable as a normal directory structure. I was thinking it was just a blob that you had to roll back to in order to get anything out of it. Snapshots appear in a .zfs/snapshot/SNAPSHOT_NAME directory at each mountpoint and are accessible as a normal read-only directory hierarchy below there. OTOH, the send stream _is_ a blob. Am I correct in assuming that one could: # zfs send -R snapshot | dd obs=10240 of=/dev/rst0 to archive it to tape instead of another [system:]drive? Yes. The output from zfs send is a stream of bytes that you can treat as you would any other stream of bytes. But this approach isn't recommended. -- Peter Jeremy pgp61ijyBCuu8.pgp Description: PGP signature
Re: Musings on ZFS Backup strategies
On 3/2/2013 4:14 PM, Peter Jeremy wrote: On 2013-Mar-01 08:24:53 -0600, Karl Denninger k...@denninger.net wrote: If I then restore the base and snapshot, I get back to where I was when the latest snapshot was taken. I don't need to keep the incremental snapshot for longer than it takes to zfs send it, so I can do: zfs snapshot pool/some-filesystem@unique-label zfs send -i pool/some-filesystem@base pool/some-filesystem@unique-label zfs destroy pool/some-filesystem@unique-label and that seems to work (and restore) just fine. This gives you an incremental since the base snapshot - which will probably grow in size over time. If you are storing the ZFS send streams on (eg) tape, rather than receiving them, you probably still want the Towers of Hanoi style backup hierarchy to control your backup volume. It's also worth noting that whilst the stream will contain the compression attributes of the filesystem(s) in it, the actual data is the stream in uncompressed I noted that. The script I wrote to do this looks at the compression status in the filesystem and, if enabled, pipes the data stream through pbzip2 on the way to storage. The only problem with this presumption is that for database data filesystems the best practices say that you should set the recordsize to that of the underlying page size of the dbms (e.g. 8k for Postgresql) for best performance and NOT enable compression. Reality however is that the on-disk format of most database files is EXTREMELY compressible (often WELL better than 2:1), so I sacrifice there. I think the better option is to stuff a user parameter into the filesystem attribute table (which apparently I can do without boundary) telling the script whether or not to compress on output so it's not tied to the filesystem's compression setting. I'm quite-curious, in fact, as to whether the best practices really are in today's world. Specifically, for a CPU-laden machine with lots of compute power I wonder if enabling compression on the database filesystems and leaving the recordsize alone would be a net performance win due to the reduction in actual I/O volume. This assumes you have the CPU available, of course, but that has gotten cheaper much faster than I/O bandwidth has. This in turn means that keeping more than two incremental dumps offline has little or no value; the second merely being taken to insure that there is always at least one that has been written to completion without error to apply on top of the base. This is quite a critical point with this style of backup: The ZFS send stream is not intended as an archive format. It includes error detection but no error correction and any error in a stream renders the whole stream unusable (you can't retrieve only part of a stream). If you go this way, you probably want to wrap the stream in a FEC container (eg based on ports/comms/libfec) and/or keep multiple copies. That's no more of a problem than it is for a dump file saved on a disk though, is it? While restore can (putatively) read past errors on a tape, in reality if the storage is a disk and part of the file is unreadable the REST of that particular archive is unreadable. Skipping unreadable records does sorta work for tapes, but it rarely if ever does for storage onto a spinning device within the boundary of the impacted file. In practice I attempt to cover this by (1) saving the stream to local disk and then (2) rsync'ing the first disk to a second in the same cabinet. If the file I just wrote is unreadable I should discover it at (2), which hopefully is well before I actually need it in anger. Disk #2 then gets rotated out to an offsite vault on a regular schedule in case the building catches fire or similar. My exposure here is to time-related bitrot which is a non-zero risk but I can't scrub a disk that's sitting in a vault, so I don't know that there's a realistic means around this risk other than a full online hotsite that I can ship the snapshots to (which I don't have the necessary bandwidth or storage to cover.) If I change the backup media (currently UFS formatted) to ZFS formatted and dump directly there via a zfs send/receive I could run both drives as a mirror instead of rsync'ing from one to the other after the first copy is done, then detach the mirror to rotate the drive out and attach the other one, causing a resilver. That's fine EXCEPT if I have a controller go insane I now probably lose everything other than the offsite copy since everything is up for write during the snapshot operation. That ain't so good and that's a risk I've had turn into reality twice in 20 years. On the upside if the primary has an error on it I catch it when I try to resilver as that operation will fail since the entire data structure that's on-disk and written has to be traversed and the checksums should catch any silent corruption. If that happens I know I'm naked (other than the vault copy which I hope is good!) until I replace the
Re: Musings on ZFS Backup strategies
- Original Message - From: Karl Denninger k...@denninger.net Reality however is that the on-disk format of most database files is EXTREMELY compressible (often WELL better than 2:1), so I sacrifice there. I think the better option is to stuff a user parameter into the filesystem attribute table (which apparently I can do without boundary) telling the script whether or not to compress on output so it's not tied to the filesystem's compression setting. I'm quite-curious, in fact, as to whether the best practices really are in today's world. Specifically, for a CPU-laden machine with lots of compute power I wonder if enabling compression on the database filesystems and leaving the recordsize alone would be a net performance win due to the reduction in actual I/O volume. This assumes you have the CPU available, of course, but that has gotten cheaper much faster than I/O bandwidth has. We've been using ZFS compression on mysql filesystems for quite some time and have good success with it. It is dependent on the HW as you say though so you need to know where the bottleneck is in your system, cpu or disk. mysql 5.6 also added better recordsize support which could be interesting. Also be aware of the additional latency the compression can add. I'm also not 100% sure that the compression in ZFS scales beyond one core its been something I've meant to look in to / test but not got round to. Regards Steve This e.mail is private and confidential between Multiplay (UK) Ltd. and the person or entity to whom it is addressed. In the event of misdirection, the recipient is prohibited from using, copying, printing or otherwise disseminating it or any information contained in it. In the event of misdirection, illegible or incomplete transmission please telephone +44 845 868 1337 or return the E.mail to postmas...@multiplay.co.uk. ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to freebsd-stable-unsubscr...@freebsd.org
Re: Musings on ZFS Backup strategies
The recommended approach is to do zfs send | zfs recv and store a replica of your pool (with whatever level of RAID that meets your needs). This way, you immediately detect an error in the send stream and can repeat the send. You then use scrub to verify (and recover) the replica. I do zfs send | zfs recv from several machines to a backup server in a different building. Each day an incremental send is done using the previous day's incremental send as the base. One reason for this approach is to minimize the amount of bandwidth required since one of the machines is across a T1. This technique requires keeping a record of the current base snapshot for each filesystem, and a system in place to keep from destroying the base snapshot. I learned the latter the hard way when a machine went down for several days, and when it came back up the script that destroys out-of-date snapshots deleted the incremental base snapshot. I'm running 9.1-stable with zpool features on my machines, and with this upgrade came zfs hold and zfs release. This allows you to lock a snapshot so it can't be destroyed until it's released. With this feature, I do the following for each filesystem: zfs send -i yesterdays_snapshot todays_snapshot | ssh backup_server zfs recv on success: zfs hold todays_snapshot zfs release yesterdays_snapshot ssh backup_server zfs hold todays_snapshot ssh backup_server zfs release yesterdays_snapshot update zfs_send_dates file with filesystem and snapshot name John Theus TheUsGroup.com ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to freebsd-stable-unsubscr...@freebsd.org
Re: Musings on ZFS Backup strategies
Quoth Ben Morrow: I don't know what medium you're backing up to (does anyone use tape any more?) but when backing up to disk I much prefer to keep the backup in the form of a filesystem rather than as 'zfs send' streams. One reason for this is that I believe that new versions of the ZFS code are more likely to be able to correctly read old versions of the filesystem than old versions of the stream format; this may not be correct any more, though. Another reason is that it means I can do 'rolling snapshot' backups. I do an initial dump like this # zpool is my working pool # bakpool is a second pool I am backing up to zfs snapshot -r zpool/fs at dump http://lists.freebsd.org/mailman/listinfo/freebsd-stable zfs send -R zpool/fs at dump http://lists.freebsd.org/mailman/listinfo/freebsd-stable | zfs recv -vFd bakpool That pipe can obviously go through ssh or whatever to put the backup on a different machine. Then to make an increment I roll forward the snapshot like this zfs rename -r zpool/fs at dump http://lists.freebsd.org/mailman/listinfo/freebsd-stable dump-old zfs snapshot -r zpool/fs at dump http://lists.freebsd.org/mailman/listinfo/freebsd-stable zfs send -R -I @dump-old zpool/fs at dump http://lists.freebsd.org/mailman/listinfo/freebsd-stable | zfs recv -vFd bakpool zfs destroy -r zpool/fs at dump-old http://lists.freebsd.org/mailman/listinfo/freebsd-stable zfs destroy -r bakpool/fs at dump-old http://lists.freebsd.org/mailman/listinfo/freebsd-stable (Notice that the increment starts at a snapshot called @dump-old on the send side but at a snapshot called @dump on the recv side. ZFS can handle this perfectly well, since it identifies snapshots by UUID, and will rename the bakpool snapshot as part of the recv.) This brings the filesystem on bakpool up to date with the filesystem on zpool, including all snapshots, but never creates an increment with more than one backup interval's worth of data in. If you want to keep more history on the backup pool than the source pool, you can hold off on destroying the old snapshots, and instead rename them to something unique. (Of course, you could always give them unique names to start with, but I find it more convenient not to.) Uh, I see a potential problem here. What if the zfs send | zfs recv command fails for some reason before completion? I have noted that zfs recv is atomic -- if it fails for any reason the entire receive is rolled back like it never happened. But you then destroy the old snapshot, and the next time this runs the new gets rolled down. It would appear that there's an increment missing, never to be seen again. What gets lost in that circumstance? Anything changed between the two times -- and silently at that? (yikes!) -- -- Karl Denninger /The Market Ticker ®/ http://market-ticker.org Cuda Systems LLC ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to freebsd-stable-unsubscr...@freebsd.org
Re: Musings on ZFS Backup strategies
Quoth Karl Denninger k...@denninger.net: Quoth Ben Morrow: I don't know what medium you're backing up to (does anyone use tape any more?) but when backing up to disk I much prefer to keep the backup in the form of a filesystem rather than as 'zfs send' streams. One reason for this is that I believe that new versions of the ZFS code are more likely to be able to correctly read old versions of the filesystem than old versions of the stream format; this may not be correct any more, though. Another reason is that it means I can do 'rolling snapshot' backups. I do an initial dump like this # zpool is my working pool # bakpool is a second pool I am backing up to zfs snapshot -r zpool/fs at dump zfs send -R zpool/fs at dump | zfs recv -vFd bakpool That pipe can obviously go through ssh or whatever to put the backup on a different machine. Then to make an increment I roll forward the snapshot like this zfs rename -r zpool/fs at dump dump-old zfs snapshot -r zpool/fs at dump zfs send -R -I @dump-old zpool/fs at dump | zfs recv -vFd bakpool zfs destroy -r zpool/fs at dump-old zfs destroy -r bakpool/fs at dump-old (Notice that the increment starts at a snapshot called @dump-old on the send side but at a snapshot called @dump on the recv side. ZFS can handle this perfectly well, since it identifies snapshots by UUID, and will rename the bakpool snapshot as part of the recv.) This brings the filesystem on bakpool up to date with the filesystem on zpool, including all snapshots, but never creates an increment with more than one backup interval's worth of data in. If you want to keep more history on the backup pool than the source pool, you can hold off on destroying the old snapshots, and instead rename them to something unique. (Of course, you could always give them unique names to start with, but I find it more convenient not to.) Uh, I see a potential problem here. What if the zfs send | zfs recv command fails for some reason before completion? I have noted that zfs recv is atomic -- if it fails for any reason the entire receive is rolled back like it never happened. But you then destroy the old snapshot, and the next time this runs the new gets rolled down. It would appear that there's an increment missing, never to be seen again. No, if the recv fails my backup script aborts and doesn't delete the old snapshot. Cleanup then means removing the new snapshot and renaming the old back on the source zpool; in my case I do this by hand, but it could be automated given enough thought. (The names of the snapshots on the backup pool don't matter; they will be cleaned up by the next successful recv.) What gets lost in that circumstance? Anything changed between the two times -- and silently at that? (yikes!) It's impossible to recv an incremental stream on top of the wrong snapshot (identified by UUID, not by its current name), so nothing can get silently lost. A 'zfs recv -F' will find the correct starting snapshot on the destination filesystem (assuming it's there) regardless of its name, and roll forward to the state as of the end snapshot. If a recv succeeds you can be sure nothing up to that point has been missed. The worst that can happen is if you mistakenly delete the snapshot on the source pool that marks the end of the last successful recv on the backup pool; in that case you have to take an increment from further back (which will therefore be a larger incremental stream than it needed to be). The very worst case is if you end up without any snapshots in common between the source and backup pools, and you have to start again with a full dump. Ben ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to freebsd-stable-unsubscr...@freebsd.org
Re: Musings on ZFS Backup strategies
On 3/2/2013 10:23 PM, Ben Morrow wrote: Quoth Karl Denninger k...@denninger.net: Quoth Ben Morrow: I don't know what medium you're backing up to (does anyone use tape any more?) but when backing up to disk I much prefer to keep the backup in the form of a filesystem rather than as 'zfs send' streams. One reason for this is that I believe that new versions of the ZFS code are more likely to be able to correctly read old versions of the filesystem than old versions of the stream format; this may not be correct any more, though. Another reason is that it means I can do 'rolling snapshot' backups. I do an initial dump like this # zpool is my working pool # bakpool is a second pool I am backing up to zfs snapshot -r zpool/fs at dump zfs send -R zpool/fs at dump | zfs recv -vFd bakpool That pipe can obviously go through ssh or whatever to put the backup on a different machine. Then to make an increment I roll forward the snapshot like this zfs rename -r zpool/fs at dump dump-old zfs snapshot -r zpool/fs at dump zfs send -R -I @dump-old zpool/fs at dump | zfs recv -vFd bakpool zfs destroy -r zpool/fs at dump-old zfs destroy -r bakpool/fs at dump-old (Notice that the increment starts at a snapshot called @dump-old on the send side but at a snapshot called @dump on the recv side. ZFS can handle this perfectly well, since it identifies snapshots by UUID, and will rename the bakpool snapshot as part of the recv.) This brings the filesystem on bakpool up to date with the filesystem on zpool, including all snapshots, but never creates an increment with more than one backup interval's worth of data in. If you want to keep more history on the backup pool than the source pool, you can hold off on destroying the old snapshots, and instead rename them to something unique. (Of course, you could always give them unique names to start with, but I find it more convenient not to.) Uh, I see a potential problem here. What if the zfs send | zfs recv command fails for some reason before completion? I have noted that zfs recv is atomic -- if it fails for any reason the entire receive is rolled back like it never happened. But you then destroy the old snapshot, and the next time this runs the new gets rolled down. It would appear that there's an increment missing, never to be seen again. No, if the recv fails my backup script aborts and doesn't delete the old snapshot. Cleanup then means removing the new snapshot and renaming the old back on the source zpool; in my case I do this by hand, but it could be automated given enough thought. (The names of the snapshots on the backup pool don't matter; they will be cleaned up by the next successful recv.) I was concerned that if the one you rolled to old get killed without the backup being successful then you're screwed as you've lost the context. I presume that zfs recv will properly set the exit code non-zero if something's wrong (I would hope so!) What gets lost in that circumstance? Anything changed between the two times -- and silently at that? (yikes!) It's impossible to recv an incremental stream on top of the wrong snapshot (identified by UUID, not by its current name), so nothing can get silently lost. A 'zfs recv -F' will find the correct starting snapshot on the destination filesystem (assuming it's there) regardless of its name, and roll forward to the state as of the end snapshot. If a recv succeeds you can be sure nothing up to that point has been missed. Ah, ok. THAT I did not understand. So the zfs recv process checks what it's about to apply the delta against, and if it can't find a consistent place to start it garfs rather than screw you. That's good. As long as it gets caught I can live with it. Recovery isn't a terrible pain in the butt so long as it CAN be recovered. It's the potential for silent failures that scare the bejeezus out of me for all the obvious reasons. The worst that can happen is if you mistakenly delete the snapshot on the source pool that marks the end of the last successful recv on the backup pool; in that case you have to take an increment from further back (which will therefore be a larger incremental stream than it needed to be). The very worst case is if you end up without any snapshots in common between the source and backup pools, and you have to start again with a full dump. Ben Got it. That's not great in that it could force a new full copy, but it's also not the end of the world. In my case I am already automatically taking daily and 4-hour snaps, keeping a week's worth around, which is more than enough time to be able to obtain a consistent place to go from. That should be ok then. I think I'm going to play with this and see what I think of it. One thing that is very attractive to this design is to have the receiving side be a mirror, then to rotate to the vault copy run a scrub (to insure that both members are
Re: Musings on ZFS Backup strategies
Karl Denninger (karl) writes: I think I'm going to play with this and see what I think of it. One thing that is very attractive to this design is to have the receiving side be a mirror, then to rotate to the vault copy run a scrub (to insure that both members are consistent at a checksum level), break the mirror and put one in the vault, replacing it with the drive coming FROM the vault, then do a zpool replace and allow it to resilver into the other drive. You now have the two in consistent state again locally if the pool pukes and one in the vault in the event of a fire or other entire facility is toast event. That's one solution. The only risk that makes me uncomfortable doing this is that the pool is always active when the system is running. With UFS backup disks it's not -- except when being actually written to they're unmounted, and this materially decreases the risk of an insane adapter scribbling the drives, since there is no I/O at all going to them unless mounted. While the backup pool would be nominally idle it is probably more-exposed to a potential scribble than the UFS-mounted packs would be. Could zpool export in between syncs on the target, assuming that's not your root pool :) Cheers, Phil ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to freebsd-stable-unsubscr...@freebsd.org
Re: Musings on ZFS Backup strategies
Quoth Phil Regnauld regna...@x0.dk: The only risk that makes me uncomfortable doing this is that the pool is always active when the system is running. With UFS backup disks it's not -- except when being actually written to they're unmounted, and this materially decreases the risk of an insane adapter scribbling the drives, since there is no I/O at all going to them unless mounted. While the backup pool would be nominally idle it is probably more-exposed to a potential scribble than the UFS-mounted packs would be. Could zpool export in between syncs on the target, assuming that's not your root pool :) If I were feeling paranoid I might be tempted to not only keep the pool exported when not in use, but to 'zpool offline' one half of the mirror while performing the receive, then put it back online and allow it to resilver before exporting the whole pool again. I'm not sure if there's any way to wait for the resilver to finish except to poll 'zpool status', though. Ben ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to freebsd-stable-unsubscr...@freebsd.org