Hail license change was just pushed to the github hail repository.
Jeff
--
To unsubscribe from this list: send the line unsubscribe hail-devel in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
(CC'd hail-devel list)
On 11/26/2012 02:28 AM, Hideki Yamane wrote:
Hello hail upstream authors,
I'm interested in porting hail (and Aeolus) to Debian, but have some
questions about it.
Cool!
Q: Is this project is still alive? if so, where is the current main site.
Could you tell
kernel.org fixed their wiki system, which means that all the k.org wikis
are once again read-write! This includes Project Hail's home page,
https://hail.wiki.kernel.org/
I hope to have the git repos moved back from https://github.com/jgarzik/
to kernel.org soon also.
Jeff
--
On 01/25/2012 08:40 PM, Pete Zaitcev wrote:
Jeff, looks like the wiki rots. The login points to this URL
https://hail.wiki.kernel.org/articles/u/s/e/Special%7EUserLogin_94cd.html
It returns 404. HALP?
Yes -- all kernel.org wikis are _still_ read-only, even this many months
after the
On 10/14/2011 01:34 PM, Pete Zaitcev wrote:
Signed-off-by: Pete Zaitcevzait...@kotori.zaitcev.us
---
lib/hstor.c |2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/lib/hstor.c b/lib/hstor.c
index cb9c4da..5ce9b76 100644
--- a/lib/hstor.c
+++ b/lib/hstor.c
@@ -761,7 +761,7
On 10/04/2011 12:12 PM, Pete Zaitcev wrote:
Are we going to have a git tree somewhere? It looks like our old one
was purged from git.kernel.org.
Sorry, I should have posted. It was migrated along with the kernel.org
trees to
https://github.com/jgarzik/{hail,tabled,itd,nfs4d}
On 05/05/2011 10:14 AM, Jim Meyering wrote:
FYI, /usr/include/rpc/ no longer exists, as of F15's glibc-headers-2.13.90-10,
so hail's lib/cld_msg_rpc.h will have to do something about this #include:
$ grep rpc.h lib/cld_msg_rpc.h
#includerpc/rpc.h
hm. Surely they did not delete
I just pushed the CLD protocol change (UDP - TCP) to hail.git[1]. See
the original post[2] for more details. It seems pretty solid from my
beating on it, but it's still raw code.
The focus will be on hammering out the kinks in this switch over the
next 7-10 days, so expect some breakage
On 01/02/2011 06:32 PM, Pete Zaitcev wrote:
On Fri, 31 Dec 2010 05:57:28 -0500
Jeff Garzikj...@garzik.org wrote:
+ struct cldc_tcp *tcp = private;
+ ssize_t rc;
+ struct ubbp_header ubbp;
+
+ memcpy(ubbp.magic, CLD1, 4);
+ ubbp.op_size = (buflen 8) | 1;
+#ifdef
On 01/02/2011 08:20 PM, Pete Zaitcev wrote:
Looks like Rawhide throws this if libdb-devel is in use:
make check-TESTS
make[3]: Entering directory `/q/zaitcev/hail/hail-tip/test/cld'
PASS: prep-db
DB_ENV-lsn_reset: method not permitted before handle's open method
DB_ENV-dbremove: method not
Switch CLD from hand-rolled server poll code, to libevent. Follows
similar techniques and rationale as chunkd commit
c1aed7464f237e5a6309351bf003162c77d69e27. This reverts ancient commit
90b3b5edcf5aa00577f4395fdbb490ed7e9be824.
Signed-off-by: Jeff Garzik jgar...@redhat.com
---
cld
to TCP promises to make the current codebase
much easier to use, while avoiding the reinvent TCP, by using UDP
problem, which was a rabbit hole threatening CLD.
Signed-off-by: Jeff Garzik jgar...@redhat.com
---
chunkd/cldu.c|6
cld/cld.h| 43 ++
cld/msg.c
On 11/28/2010 08:41 PM, Pete Zaitcev wrote:
This patch adds the first new back-end and makes some changes to the way
nodes are added, to make the invariants of storage_node more sensible.
The filesystem back-end itself is not intended for production use,
so it makes no attempt to run any
On 11/28/2010 08:41 PM, Pete Zaitcev wrote:
This patch allows to use tabled with OpenStack Swift object store as if it
were our chunkserver, with some extra tricks. The configuration has to be
entred manually into CLD, just like in case of filesystem back-end.
The code is fairly experimental,
On 12/05/2010 10:53 PM, Pete Zaitcev wrote:
Amazon appears to give up on forcing users to migrate and bucket-in-path
format is going to stay. However, they still refuse to list buckets from
other regions on the default endpoint, which leads to annoying indirection
(need to know the region
On 11/28/2010 08:39 PM, Pete Zaitcev wrote:
This helps copy-paste safer later, mostly.
Signed-off-by: Pete Zaitcevzait...@redhat.com
---
server/object.c |2 -
server/storage.c | 79 ++---
server/tabled.h | 12 +++---
3 files changed, 53
On 12/05/2010 04:56 PM, Pete Zaitcev wrote:
Autoconf printed a warning when reconfiguting Hail, so I gave up and
added this:
[...]
Now I have a directory m4/ with symlinks... This does not seem to be
helping any portability, unless I miss where the promised macro are
being saved locally. What
On 12/06/2010 12:44 PM, Pete Zaitcev wrote:
On Mon, 06 Dec 2010 12:32:22 -0500
Jeff Garzikj...@garzik.org wrote:
Keeping the correct libtool macros in-tree implies adding a pointless
maintenance burden. The distro always gives us correct, up-to-date
files. Why would hail want to potentially
On 10/26/2010 03:47 PM, Pete Zaitcev wrote:
Now that we have a common library for Hail, an opportunity opens to trim
some duplication, such as stc_readport. It even had a comment about it.
Note that we leave cld_readport in the API for a few weeks, while I get
my tabled trees and RPMs in order.
Just committed this:
commit 57c4be44cdfa6c0cda6cf26d19e8048a945c5a78
Author: Jeff Garzik j...@garzik.org
Date: Sat Oct 23 14:01:20 2010 -0400
Use libhail's atcp rather than our own async TCP write code.
Should be functionally equivalent, as atcp originated from tabled
code
On 10/20/2010 04:53 AM, Jim Meyering wrote:
Jeff Garzik wrote:
...
Hi Jeff.
Sorry I didn't notice that the first time.
I built with ./autogen.sh ./configure make.
It looks like you recommend -Wall -Wshadow.
The two warnings above are the only ones I see with the patch,
and they're easy
.
Git shortlog attached.
Jeff Garzik (12):
chunkd: Add checksum table to on-disk format, one sum per 64k of data
chunkd: checksum data prior to returning via GET
chunk: Add Get-Partial-Object (GET_PART) operation
lib/chunksrv.c: add FIXME
chunkd: internal 32/64-bit
Home: https://hail.wiki.kernel.org/
Git: git://git.kernel.org/pub/scm/daemon/distsrv/tabled.git
Download: http://www.kernel.org/pub/software/network/distsrv/tabled/
Version 0.5.2 release notes (NEWS):
- Permit randomly allocated TCP port, for db4 replication master
- Install etc.tabled.conf as
On 10/22/2010 11:39 PM, Jeff Garzik wrote:
Home: https://hail.wiki.kernel.org/
Git: git://git.kernel.org/pub/scm/daemon/distsrv/tabled.git
Download: http://www.kernel.org/pub/software/network/distsrv/tabled/
Version 0.5.2 release notes (NEWS):
- Permit randomly allocated TCP port, for db4
On 10/22/2010 11:22 PM, Jeff Garzik wrote:
Home: https://hail.wiki.kernel.org/
Git: git://git.kernel.org/pub/scm/daemon/distsrv/hail.git
Download: http://www.kernel.org/pub/software/network/distsrv/hail/
It seems that kernel.org mirroring is broken or extremely slow at the
moment
On 10/20/2010 04:00 AM, Jim Meyering wrote:
Jeff Garzik wrote:
On 10/06/2010 08:07 AM, Jim Meyering wrote:
Make write_cb callback's buffer parameter const, like all write-like
functions.
Give a few char * parameters the const attribute.
Signed-off-by: Jim Meyeringmeyer...@redhat.com
On 10/06/2010 08:07 AM, Jim Meyering wrote:
Make write_cb callback's buffer parameter const, like all write-like
functions.
Give a few char * parameters the const attribute.
Signed-off-by: Jim Meyeringmeyer...@redhat.com
---
It looks like most of hail's interfaces are const-correct,
but one
On 09/30/2010 04:55 AM, Geert Jansen wrote:
is it correct that CLD is basically single-master right now? I can't
find any trace of the mentioned Paxos implementation in the source.
The current main branch is single-master, correct. The 'replica' branch
of hail.git contains the multi-node
On 09/29/2010 11:20 AM, Jim Meyering wrote:
chk_list_objs called fs_list_objs_open without also calling
fs_list_objs_close.
32,808 bytes in 1 blocks are definitely lost in loss record 413 of 419
at 0x4A0515D: malloc (vg_replace_malloc.c:195)
by 0x31BA8A26D0: __alloc_dir
On 09/29/2010 11:20 AM, Jim Meyering wrote:
chk_list_objs called fs_list_objs_open without also calling
fs_list_objs_close.
32,808 bytes in 1 blocks are definitely lost in loss record 413 of 419
at 0x4A0515D: malloc (vg_replace_malloc.c:195)
by 0x31BA8A26D0: __alloc_dir
For a nice code savings...
chunkd/Makefile.am |1
chunkd/chunkd.h| 28 +
chunkd/cldu.c | 64 +--
chunkd/server.c| 289 +
chunkd/util.c | 23
configure.ac |3
6 files changed, 116
On Wed, Sep 29, 2010 at 7:09 PM, Pete Zaitcev zait...@redhat.com wrote:
An interesting question is what to do when iwhd exits. I decided not to
kill what was started. So, we have a little self-contained cell of
tabled, chunkd, S3, based off a certain local directory or other
namespace anchor.
On 09/27/2010 04:53 AM, Jim Meyering wrote:
Signed-off-by: Jim Meyeringmeyer...@redhat.com
---
I would have preferred to insert a single line right before the
huri_field_escape call:
char *v = strdup(val);
[would result in a more compact, single-hunk patch]
but it looks like hail uses
On 09/27/2010 12:29 PM, Pete Zaitcev wrote:
On Mon, 27 Sep 2010 10:53:06 +0200
Jim Meyeringj...@meyering.net wrote:
- stmp = huri_field_escape(strdup(val), QUERY_ESCAPE_MASK);
+ v = strdup(val);
+ stmp = huri_field_escape(v, QUERY_ESCAPE_MASK);
str =
On 09/27/2010 08:49 PM, Pete Zaitcev wrote:
Premature optimization is the root of all evil.
Use a sensible convention of not screwing with the argument, at the expense
of extra strdup.
Fortunately, all users are confined to Hail itself, even if huri_field_escape
is exported.
Signed-off-by:
On 09/27/2010 08:52 PM, Pete Zaitcev wrote:
Our current tests do not invoke hstor_keys at all, and so they did not catch
a crash with double free in append_qparam.
Add a very basic test which at least calls hstor_keys to verify that it
does not crash right away. This test does not excercise
On 09/24/2010 07:32 AM, Jim Meyering wrote:
You can pull from the oom branch here:
git://git.infradead.org/users/meyering/tabled.git
Got nearly everything perfect. Need one more minor yet important
change. As described in doc/contributions.txt, every changeset MUST
have a Signed-off-by
On 09/24/2010 01:43 PM, Jim Meyering wrote:
Jeff Garzik wrote:
On 09/24/2010 07:32 AM, Jim Meyering wrote:
You can pull from the oom branch here:
git://git.infradead.org/users/meyering/tabled.git
Got nearly everything perfect. Need one more minor yet important
change. As described
On 09/23/2010 03:55 AM, Jim Meyering wrote:
Better safe than sorry...
Unreported write failures can be unpleasant.
I fixed the one below so that a failure indication
can propagate up the call tree. You might also want to
report the failure to stderr.
I let my editor automatically update the
Changes from v1:
- avoid referencing dead struct client (grep for 'invalidate_cli'),
by changing FSM callback prototype.
- insert 'void *priv' member into struct atcp_wr_state, and replace
cb_data1/cb_data2 callback parameters with (struct atcp_wr_state *, void *).
struct client / struct
On 09/23/2010 11:28 AM, Jim Meyering wrote:
Every developer should have MALLOC_PERTURB_=N (N in 1..255) set in
his/her environment on glibc-based systems. Almost all the time.
I heard about it a while ago, even submitted a bugzilla bug to have it
documented adequately. But apparently its
On 09/22/2010 10:37 PM, Pete Zaitcev wrote:
On Wed, 22 Sep 2010 21:26:13 -0400
Jeff Garzikj...@garzik.org wrote:
It is a common idiom even in GLib that callbacks receive two anonymous
pointers; witness the data type GFunc's 'data' and 'user_data'
arguments:
On 09/22/2010 10:37 PM, Pete Zaitcev wrote:
On Wed, 22 Sep 2010 21:26:13 -0400
Jeff Garzikj...@garzik.org wrote:
So, we go a longer route and re-hook the list of completions
to a per-server global instead of a client. The patch is straight-
forward. The only thing we need to
On Tue, Sep 21, 2010 at 5:06 PM, Steven Dake sd...@redhat.com wrote:
libevent version 2 has proper mutual exclusion, but the code needs some
work.
1.x should work for chunkd at the moment. I need to resist my own
urge to think too far ahead and overengineer for the future sometimes;
I think
Just pushed this out to hail.git.
--
To unsubscribe from this list: send the line unsubscribe hail-devel in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
On 09/10/2010 08:55 AM, Jim Meyering wrote:
* server/msg.c (msg_get): Copy only name_len bytes, then NUL-terminate,
rather than using snprintf to copy up to and including nonexistent NUL.
---
valgrind exposed this. The use of snprintf would have been
correct if the inode name buffer
This patchset is just about ready to go upstream. Just need to write a
couple tests (familiar refrain eh?:)).
These changes add a new Get-Partial-Object (GET_PART) chunkd operation.
GET_PART permits partial retrieval of an object, by adding an
(offset,length) pair to the standard Get-Object
commit f1de17a6e2b3afdbfbfa581228280b65a4a17e5f
Author: Jeff Garzik j...@garzik.org
Date: Thu Aug 5 17:47:03 2010 -0400
chunkd: Add checksum table to on-disk format, one sum per 64k of data
Signed-off-by: Jeff Garzik jgar...@redhat.com
chunkd/be-fs.c | 162
On 08/12/2010 03:22 PM, Pete Zaitcev wrote:
Allow random ports for replication master to listen on.
The patch is somewhat larger than expected, because before we had
the MASTER file written right after locking. Now we may have it
written without listening parameters, and the slaves must be
On 08/05/2010 11:40 PM, Pete Zaitcev wrote:
Signed-off-by: Pete Zaitcevzait...@redhat.com
---
server/server.c |2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
commit 93b990f68e5c2c652759a2db8af049d172b8489c
Author: Pete Zaitcevzait...@yahoo.com
Date: Thu Aug 5 20:33:21 2010 -0600
On 08/04/2010 07:16 PM, Pete Zaitcev wrote:
Testing found that tabled and chunkd running on CPUs with different
word length cannot talk to each other.
The bug was introduced by commit ea5d20bc22aeed077312c9c1824e84651af17a16.
The fix is to add named padding that takes the place of the
On 07/29/2010 01:41 PM, Pete Zaitcev wrote:
On Tue, 20 Jul 2010 16:34:19 -0400
Jeff Garzikj...@garzik.org wrote:
lib/hstor.c | 147 +++---
1 file changed, 104 insertions(+), 43 deletions(-)
applied
It's not in the git repo. Check this URL:
On 07/29/2010 10:49 PM, Pete Zaitcev wrote:
Signed-off-by: Pete Zaitcevzait...@redhat.com
---
configure.ac |2 --
1 file changed, 2 deletions(-)
commit 00be6055a3801ef8e84a4c78b43b43b67a76eab9
Author: Pete Zaitcevzait...@yahoo.com
Date: Thu Jul 29 19:10:05 2010 -0600
Drop
On 07/20/2010 04:16 PM, Pete Zaitcev wrote:
Some of my performance tests for tabled hit truncation again:
[zait...@hitlain tests]$ ./poke5 -v -h niphredil.zaitcev.lan -u auser
-p apass -b test -o -k testkey-hitlain/73b84a11e6d83c65e45853338d646042
-f
Note that we are checksumming hot cache data, so SHA1 isn't as
punishing as one might think.
chunkd/be-fs.c | 51 ++-
1 file changed, 50 insertions(+), 1 deletion(-)
commit 2211e3b58620093866be4130397cb3b476620725
Author: Jeff Garzik j
On 07/17/2010 11:45 PM, Steven Dake wrote:
On 07/16/2010 10:46 PM, Jeff Garzik wrote:
chunkd: remove sendfile(2) zero-copy support
chunkd will be soon checksumming data in main memory. That removes
the utility of a zero-copy interface which bypasses the on-heap
data requirement.
Signed-off
chunkd/be-fs.c | 145 +---
chunkd/chunkd.h |3 +
2 files changed, 131 insertions(+), 17 deletions(-)
commit 394109d5c2fc2d15d91c2d36eecd57594922c1b3
Author: Jeff Garzik j...@garzik.org
Date: Sat Jul 17 01:05:15 2010 -0400
chunkd
This patchset is part of the work necessary to get ranged-GET (aka
partial GET) working. As explained in
http://marc.info/?l=hail-develm=127871407125539w=2 the current chunkd
checksum scheme does not work at all for partial retrievals, and must be
revamped.
These patches present step 1 of 4,
commit d663521ba7e6a808be02633e57dbeb7a95973c0f
Author: Jeff Garzik j...@garzik.org
Date: Thu Jul 15 13:50:10 2010 -0400
chunkd: remove sendfile(2) zero-copy support
chunkd will be soon checksumming data in main memory. That removes
the utility of a zero-copy interface which
commit e6fcc02bea062af291148771a59ee2028ae98834
Author: Jeff Garzik j...@garzik.org
Date: Thu Jul 15 13:57:17 2010 -0400
chunkd: Add checksum table to on-disk format, one sum per 64k of data
Signed-off-by: Jeff Garzik jgar...@redhat.com
chunkd/be-fs.c | 145
the following out to hail.git. If people disagree with
naming, now's the time to speak up.
commit 5188f48dd3c73ce86f2bc453a326ee0bf40fd6db
Author: Jeff Garzik j...@garzik.org
Date: Wed Jul 7 02:16:28 2010 -0400
libhail: Import httpstor, httputil modules from tabled
With the following
On Wed, Jul 07, 2010 at 06:38:22PM -0400, Jeff Garzik wrote:
Just committed the following to tabled.git on my local laptop, on a side
branch. This won't be pushed onto the main tabled branch until Friday,
to give people time to convert as zaitcev suggested in the 'new hail
repository' thread
On 07/06/2010 11:17 AM, Pete Zaitcev wrote:
On Tue, 6 Jul 2010 03:24:29 -0400
Jeff Garzikj...@garzik.org wrote:
The following patch, against current hail.git, adds the CP command to
chunkd, permitting copying from object-object inside a single table.
What is it for?
Fun! :)
More
On 07/06/2010 11:17 AM, Pete Zaitcev wrote:
On Tue, 6 Jul 2010 03:24:29 -0400
Jeff Garzikj...@garzik.org wrote:
The following patch, against current hail.git, adds the CP command to
chunkd, permitting copying from object-object inside a single table.
What is it for?
Here's a real-world
Here are a few interesting things that have appeared in the S3 API since
its initial release:
1) Object versioning. All objects now uniquely identified by (key,
version) pair. API compatibility is maintained by supporting the notion
of current version.
2) Object copying. Rather than an
This patch
* adds local, intra-table copy operation to chunkd/libhail
* illustrates what files need updating, when adding a new op to chunk
* adds some 'worker' infrastructure which should help with future ops,
notably remote copy (RCP)
* should assist tabled's implementation of S3 copy
This function seems to be missing the meat. It retrieves then
disposes of a keylist.
bool stor_obj_test(struct open_chunk *cep, uint64_t key)
{
struct st_keylist *klist;
if (!cep-stc)
return false;
klist = stc_keys(cep-stc);
if (!klist)
perform the conversion at list-objects time.
Jeff
commit ea5d20bc22aeed077312c9c1824e84651af17a16
Author: Jeff Garzik j...@garzik.org
Date: Wed Jul 7 00:51:48 2010 -0400
[chunk] protocol, disk fmt: Replace ASCII checksum representation
with binary
Rather than converting SHA1
On 07/05/2010 03:13 PM, Pete Zaitcev wrote:
On Fri, 02 Jul 2010 02:59:20 -0400
Jeff Garzikj...@garzik.org wrote:
git://git.kernel.org/pub/scm/daemon/distsrv/hail.git
libhail is a single shared library binary, linking together cldc, ncld,
libtimer, and chunkdc modules. In other
Here are a few chunkd enhancements that are currently on my drawing
board, for the near term:
(CHO_xxx denotes new chunkd network protocol commands, as listed in
include/chunk_msg.h)
* CHO_SET_SERVERS:
chunkd shall maintain a per-connection buffer known as SERVER_LIST.
This chunkd command
A new git repository
git://git.kernel.org/pub/scm/daemon/distsrv/hail.git
was created, preserving the full histories of cld.git and chunkd.git.
The existing cld.git and chunkd.git repositories have been left
untouched, for now. I also have not yet updated tabled.git for this new
Version 0.7 of hail core services has been released, at the expected places:
http://www.kernel.org/pub/software/network/distsrv/hail/
ftp://ftp.kernel.org/pub/software/network/distsrv/hail/
git://git.kernel.org/pub/scm/daemon/distsrv/hail.git
This release replaces
Coinciding with hail core v0.7 release is this tabled release, v0.5, at
the usual places:
git://git.kernel.org/pub/scm/daemon/distsrv/tabled.git
http://www.kernel.org/pub/software/network/distsrv/tabled/
ftp://ftp.kernel.org/pub/software/network/distsrv/tabled/
Release notes:
- update for
On 06/30/2010 10:49 AM, Pete Zaitcev wrote:
My rule of thumb is that magic delays are evil or stupid, so I worked on
eliminating them from our scripts. However, in this case it's just not
worth it, because the result is that we have to wait way more than 100s
for several cycles of CLD timeouts
I've been thinking about a new structure for the projects, namely having
a single hail or hail-core package, that includes cld and chunkd
services, and associated client libraries inside a new libhail.
In real terms, it would look like this:
cld - hail
libcldc
On 06/24/2010 08:31 PM, Pete Zaitcev wrote:
I worked on fixing the metadata replication in tabled. There were some
difficulties in existing code, in particular the aliasing between the
hostname used to identify nodes and the hostname used in bind() for
listening was impossible to work around in
On 06/04/2010 11:27 PM, Pete Zaitcev wrote:
I heard people say they cribbed from the same Chubby paper, but
it's bollocks. It's absolutely nothing like what Chubby implies.
No locks for one thing. To be sure, Zookeeper provides a canned
piece of code which implements locks, kinda like you can
On 05/21/2010 12:54 AM, Pete Zaitcev wrote:
- host, sizeof(host), NULL, 0, NI_NUMERICHOST);
+ host, sizeof(host), port, sizeof(port), NI_NUMERICHOST);
host[sizeof(host) - 1] = 0;
- applog(LOG_INFO, client %s connected%s, host,
+
On 05/21/2010 12:54 AM, Pete Zaitcev wrote:
After a period of uptime, chunkd may stop working with this:
May 20 08:51:47 azdragon2 chunkd[4034]: tcp accept: Too many open files
An examination with lsof shows that file descriptors for sockets and
object data files are leaked in neat pairs. As
On 05/21/2010 12:54 AM, Pete Zaitcev wrote:
The upside of this cleanup is an ease of reading and evaluating with
fewer control paths.
[This patch will only work if patch 2/6 is applied. Sorry.]
Signed-off-by: Pete Zaitcevzait...@redhat.com
---
server/chunkd.h |2 +-
server/object.c |
On 05/25/2010 11:30 PM, Pete Zaitcev wrote:
If a chunkserver goes down, tabled sometimes throws a phantom object
not found. It happens because we keep hitting the same down node and
exhaust the retries. The existing code calls rand() every time and
hopes for the best, but this is too likely to
As of itd commit 196e8f317fc7202460d7adde93dac939caf23f5d, the iSCSI
target daemon appears to survive stress tests, and does not leak memory.
I call that a good first milestone.
Jeff
--
To unsubscribe from this list: send the line unsubscribe hail-devel in
the body of a message
As of commit 23a5795e3ca555a6454b199e071482bb50655508, itd is passing
integrity and stress tests from two test suites, iscsi-harness found in
netbsd-iscsi pkg, and basic blkdev integrity tests using dd(1).
There is a whopping big memory leak that needs fixing, but the basics
appear to be
On 05/03/2010 12:07 AM, Pete Zaitcev wrote:
I suspect I copy-pasted over it when I converted to ncld, but anyhow this
patch seems work and do what's expected for --user flag.
Signed-off-by: Pete Zaitcevzait...@redhat.com
---
tools/cldcli.c |2 +-
1 file changed, 1 insertion(+), 1
On 05/01/2010 12:51 AM, Pete Zaitcev wrote:
We stopped using libevent in Chunk a while ago, but for some reason
not all references were removed. I tested this patch by building
on a fresh Fedora 13 system without libevent.
Signed-off-by: Pete Zaitcevzait...@redhat.com
---
configure.ac
Hail devs,
Project Hail was, in part, conceived as an umbrella of libraries and
services enabling the mating of a well known, Internet-standard API with
a back-end that enables distributed storage. tabled is an example of
this: it provides an application front-end compatible with S3 API,
On 04/18/2010 12:42 AM, Pete Zaitcev wrote:
On Fedora 14, the following is seen in syslog:
Apr 17 19:58:52 niphredil tabled: tabled: connecting to site
hitlain.zaitcev.lan:8083: No route to host
Apr 17 19:58:56 niphredil tabled: tabled: DB_ENV-rep_elect:WARNING:
nvotes (1) is sub-majority
On 04/16/2010 10:18 PM, Pete Zaitcev wrote:
On Wed, 14 Apr 2010 15:55:01 -0400
Jeff Garzikj...@garzik.org wrote:
+++ b/lib/Makefile.am
@@ -27,6 +27,7 @@ libcldc_la_SOURCES= \
common.c\
libtimer.c \
pkt.c \
+
On 04/17/2010 09:36 PM, Pete Zaitcev wrote:
Is there a reason why the main_loop in chunkd uses a naked
g_hash_table_lookup instead of srv_poll_lookup? Performance?
@@ -1681,8 +1681,7 @@ static int main_loop(void)
fired++;
- sp =
The same source, same spec.
Build #1 (fails on x86_64):
http://koji.fedoraproject.org/koji/taskinfo?taskID=2119825
Build #2 (fails on i686):
http://koji.fedoraproject.org/koji/taskinfo?taskID=2120174
Build #3 (success on all platforms):
On 04/14/2010 02:34 PM, Pete Zaitcev wrote:
Currently, there's nothing in the verbose output about sent packets
at all. No, really! This is very confusing, even if I run tcpdump
in the same time. I think we should add this.
Signed-off-by: Pete Zaitcevzait...@redhat.com
---
lib/cldc.c |2
On 04/14/2010 02:35 PM, Pete Zaitcev wrote:
Same as everywhere else: missing prototypes, so implementations are
not actually matched by the compiler.
Signed-off-by: Pete Zaitcevzait...@redhat.com
---
lib/readport.c |1 +
test/libtest.c |1 +
2 files changed, 2 insertions(+)
On 04/14/2010 02:33 PM, Pete Zaitcev wrote:
You know what's weird... Without this, I cannot build an RPM at all,
the rpmbuild complains about unpackaged files and aborts. But
everyone else seems to have no problem? Strange. BTW, I am on
Fedora 14.
Signed-off-by: Pete Zaitcevzait...@redhat.com
On 03/31/2010 08:43 PM, Pete Zaitcev wrote:
diff --git a/server/server.c b/server/server.c
index 3208e0f..2d68ee6 100644
--- a/server/server.c
+++ b/server/server.c
@@ -55,7 +55,7 @@ static struct argp_option options[] = {
Store database environment in DIRECTORY. Default:
On 04/01/2010 09:51 PM, Pete Zaitcev wrote:
From: Jeff Garzikjgar...@pobox.com
Subject: Re: Tabled issues
Date: Mon, 29 Mar 2010 15:32:33 -0400
I asserted that the standard stats dump facility must dump
all available statistics. That does not exclude other methods
of stat(us) dumping. Your
On 04/01/2010 09:51 PM, Pete Zaitcev wrote:
At certain network and disk speeds, tabled can blow its stack by
filling it with (essentially) endless recursion:
#2 0x0040c077 in cli_write_free (cli=value optimized out, tmp=
0x7bb910, done=value optimized out) at server.c:397
#3
On 04/06/2010 11:32 PM, Pete Zaitcev wrote:
On Tue, 06 Apr 2010 10:40:33 -0400
Jeff Garzikj...@garzik.org wrote:
The debug levels are
0: key messages affecting server operation, only
1: debugging output enabled, sans per-packet output
2: debugging output enabled,
On 04/01/2010 07:01 AM, Samba - BoYang wrote:
hi, *
CLD doesn't build on db-4.3 on suse 11, since db-4.3 uses deprecated
structure members DBC-c_xxx(c_close(), etc) instead of DBC-xxx. :-)
It won't build on db-4.4, either. probably won't build on db-4.5, as
db-5.0 says DBC-xxx was
On 03/28/2010 09:57 PM, Pete Zaitcev wrote:
Signed-off-by: Pete Zaitcevzait...@redhat.com
---
server/storage.c |3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
applied
--
To unsubscribe from this list: send the line unsubscribe hail-devel in
the body of a message to
On 03/16/2010 05:59 AM, Akinobu Mita wrote:
At the beginning of stc_parse_key(), st_object is allocated twice for
the same variable.
Signed-off-by: Akinobu Mitaakinobu.m...@gmail.com
---
lib/chunkdc.c |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
good catch, applied
--
To
1 - 100 of 196 matches
Mail list logo
