You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Greg Stein <gs...@lyra.org> on 2003/06/11 19:47:37 UTC
the 0.24 release?
Michael,
The 0.24 release is scheduled for tomorrow. While we didn't happen to knock
out the items scheduled in IZ for the release, it still seems very important
to *have* one because of all the work that has gone into it. Ben points out
that we actually forked for the 0.23 release about six weeks ago.
And while Mike has got his FS work nearly ready to merge into the trunk, it
would seem appropriate to let that bake in the trunk for developers rather
than having our users be the first people to try it out. Thus, releasing
0.24 before that merge would seem to be appropriate.
I'm not aware of any major issue that needs to be fixed for the release, so
it can basically be cut any time now.
Are you available to do the release tonite / tomorrow?
Thanks,
-g
--
Greg Stein, http://www.lyra.org/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by John Peacock <jp...@rowman.com>.
Jamin W. Collins wrote:
> There are most likely a few more, but these seem to cover the ones I
> recall.
>
I also reported one, but that seemed to be mostly a self-induced problem:
http://subversion.tigris.org/servlets/ReadMsg?list=dev&msgId=259060
where I unintentionally introduced a null directory element (stupid, stupid
shell autocompletion ;~). Nonetheless, all of the paths up to that point
contained null directory elements "file:///var/svn/perl//trunk" so there was no
reason it should be able to commit *anything* if it could not commit everything.
John
--
John Peacock
Director of Information Research and Technology
Rowman & Littlefield Publishing Group
4720 Boston Way
Lanham, MD 20706
301-459-3366 x.5010
fax 301-429-5747
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by "Jamin W. Collins" <jc...@asgardsrealm.net>.
On Wed, Jun 11, 2003 at 02:09:29PM -0700, Blair Zajac wrote:
> Greg Stein wrote:
>
> > Hunh? What corruption are you referring to? I'm not aware of any
> > problems at the moment. The only one that I know that has ever
> > occurred was BDB 4.0 on the Mac OS X platform.
>
> Where have you been :) There have been reports of svn/BDB 4.1.25
> repositories getting corrupted probably once a month or so.
Yea, what he said. =) Check out the following links:
http://subversion.tigris.org/servlets/ReadMsg?list=dev&msgId=258996
http://subversion.tigris.org/servlets/ReadMsg?list=dev&msgId=251824
http://subversion.tigris.org/servlets/ReadMsg?list=dev&msgId=259894
There are most likely a few more, but these seem to cover the ones I
recall.
> Collab.net was going to put somebody onto tracking down this issue but
> that hasn't happened yet.
Sounds like this is still a problem then. I realize it would break the
release cycle, but shouldn't this be resolved before the next release?
--
Jamin W. Collins
This is the typical unix way of doing things: you string together lots
of very specific tools to accomplish larger tasks. -- Vineet Kumar
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by Blair Zajac <bl...@orcaware.com>.
Robert Pluim wrote:
>
> Greg Stein writes:
> > On Wed, Jun 11, 2003 at 02:09:29PM -0700, Blair Zajac wrote:
> > > Greg Stein wrote:
> > >...
> > > > Hunh? What corruption are you referring to? I'm not aware of
> > > > any problems at the moment. The only one that I know that has
> > > > ever occurred was BDB 4.0 on the Mac OS X platform.
> > >
> > > Where have you been :) There have been reports of svn/BDB 4.1.25
> > > repositories getting corrupted probably once a month or so.
> > >
> > > Collab.net was going to put somebody onto tracking down this
> > > issue but that hasn't happened yet.
> >
> > Eh? Is this the one that is generally typified by people using multiple
> > access paths, and then you need to run recovery every now and then?
>
> FWIW, I've seen this with my test repositories, but I'd put it down to
> my use of non-standard [1] builds of svn. Using ra_local for a while
> and then ra_dav seems to be the pattern that triggers it. Never lost
> any data though, recover has always worked.
I never use ra_local on my repos, they are always accessed by ra_dav.
Best,
Blair
--
Blair Zajac <bl...@orcaware.com>
Plots of your system's performance - http://www.orcaware.com/orca/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by Robert Pluim <rp...@bigfoot.com>.
Greg Stein writes:
> On Wed, Jun 11, 2003 at 02:09:29PM -0700, Blair Zajac wrote:
> > Greg Stein wrote:
> >...
> > > Hunh? What corruption are you referring to? I'm not aware of
> > > any problems at the moment. The only one that I know that has
> > > ever occurred was BDB 4.0 on the Mac OS X platform.
> >
> > Where have you been :) There have been reports of svn/BDB 4.1.25
> > repositories getting corrupted probably once a month or so.
> >
> > Collab.net was going to put somebody onto tracking down this
> > issue but that hasn't happened yet.
>
> Eh? Is this the one that is generally typified by people using multiple
> access paths, and then you need to run recovery every now and then?
FWIW, I've seen this with my test repositories, but I'd put it down to
my use of non-standard [1] builds of svn. Using ra_local for a while
and then ra_dav seems to be the pattern that triggers it. Never lost
any data though, recover has always worked.
> When I see the word "corruption", I think "data loss." Rather than "BDB is
> freaked out. Give it a 'lude."
Yup.
Footnotes:
[1] ie "hacked to bits by me"
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: corrupted repository (was Re: the 0.24 release?)
Posted by Colin Watson <cj...@flatline.org.uk>.
On Thu, Jun 12, 2003 at 05:24:58PM -0600, Jamin W. Collins wrote:
> On Fri, Jun 13, 2003 at 01:21:02AM +0200, lodewijk@reddwarf.xs4all.nl wrote:
> > Jamin W. Collins wrote:
> > > The system does have libdb4.0 installed on it though, and there is a
> > > somewhat round-a-bout dependancy from apache2-mpm-worker to libdb4.0 (so
> > > I may be wrong):
> > >
> > > apache2-mpm-worker
> > > - apache2-common
> > > - openssl
> > > - perl
> > > - libdb4.0
> >
> > but it doesn't seem to actually be linked in:
> >
> > # ldd /usr/sbin/apache2 | grep libdb
> > libdb-4.1.so => /usr/lib/libdb-4.1.so (0x40153000)
> > # ldd /usr/lib/apache2/modules/* | grep libdb
> > libdb-4.1.so => /usr/lib/libdb-4.1.so (0x400a3000)
>
> Right, but I'm not familiar with the internals at all. If at some point
> Perl is invoked on the server side from Apache2, there's the possibility
> that it's using the wrong BDB version, right?
perl depends on libdb4.0 due to the DB_File module. I'm fairly confident
that won't affect openssl.
Cheers,
--
Colin Watson [cjwatson@flatline.org.uk]
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: corrupted repository (was Re: the 0.24 release?)
Posted by "Jamin W. Collins" <jc...@asgardsrealm.net>.
On Fri, Jun 13, 2003 at 01:21:02AM +0200, lodewijk@reddwarf.xs4all.nl wrote:
> Jamin W. Collins wrote:
>
> > I don't believe this was the case with my corruption. I've been using
> > the Debian packages on all my systems.
>
> same here. 'changes' table flipped to recno, on a Debian machine.
>
> > The system does have libdb4.0 installed on it though, and there is a
> > somewhat round-a-bout dependancy from apache2-mpm-worker to libdb4.0 (so
> > I may be wrong):
> >
> > apache2-mpm-worker
> > - apache2-common
> > - openssl
> > - perl
> > - libdb4.0
>
> but it doesn't seem to actually be linked in:
>
> # ldd /usr/sbin/apache2 | grep libdb
> libdb-4.1.so => /usr/lib/libdb-4.1.so (0x40153000)
> # ldd /usr/lib/apache2/modules/* | grep libdb
> libdb-4.1.so => /usr/lib/libdb-4.1.so (0x400a3000)
Right, but I'm not familiar with the internals at all. If at some point
Perl is invoked on the server side from Apache2, there's the possibility
that it's using the wrong BDB version, right?
--
Jamin W. Collins
Remember, root always has a loaded gun. Don't run around with it unless
you absolutely need it. -- Vineet Kumar
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: corrupted repository (was Re: the 0.24 release?)
Posted by lo...@reddwarf.xs4all.nl.
Jamin W. Collins wrote:
> I don't believe this was the case with my corruption. I've been using
> the Debian packages on all my systems.
same here. 'changes' table flipped to recno, on a Debian machine.
> The system does have libdb4.0 installed on it though, and there is a
> somewhat round-a-bout dependancy from apache2-mpm-worker to libdb4.0 (so
> I may be wrong):
>
> apache2-mpm-worker
> - apache2-common
> - openssl
> - perl
> - libdb4.0
but it doesn't seem to actually be linked in:
# ldd /usr/sbin/apache2 | grep libdb
libdb-4.1.so => /usr/lib/libdb-4.1.so (0x40153000)
# ldd /usr/lib/apache2/modules/* | grep libdb
libdb-4.1.so => /usr/lib/libdb-4.1.so (0x400a3000)
Lodewijk
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: corrupted repository (was Re: the 0.24 release?)
Posted by "Jamin W. Collins" <jc...@asgardsrealm.net>.
On Thu, Jun 12, 2003 at 10:28:05AM +0200, Timothee Besset wrote:
> I dunno if that's related, but I've seen similar things happen to my
> repository after 0.23 upgrade. It was because of db version mismatch. I
> used to have a 'default' build of httpd (which was probably using db 4
> then), and svn 0.22 built against db 4.1
>
> When I switched to 0.23 with the same setup, things started to go bad. I
> was not aware of the db 4 in httpd / db 4.1 in mod_dav_svn till then. I
> rebuilt everything making sure httpd and svn where using db 4.1, and the
> problems went away.
I don't believe this was the case with my corruption. I've been using
the Debian packages on all my systems. Remote access to the
repositories (the only way they are accessed) is via mod_dav_svn and
Apache2. The apache version install installed on the server is Debian's
apache2-mpm-worker v2.0.46-2. While I haven't specifically checked the
linking of the binary files, the Depends for the package are:
libapr0 (>= 2.0.46), libc6 (>= 2.3.1-1), libdb4.1, libexpat1 (>=
1.95.6), libgdbmg1, libssl0.9.7, zlib1g (>= 1:1.1.4), apache2-common
(= 2.0.46-2)
As you can see, it's dependant on libdb4.1 (Debian's BDB v4.1 package).
(little more investigation)
The system does have libdb4.0 installed on it though, and there is a
somewhat round-a-bout dependancy from apache2-mpm-worker to libdb4.0 (so
I may be wrong):
apache2-mpm-worker
- apache2-common
- openssl
- perl
- libdb4.0
So, is this the root of the problem?
--
Jamin W. Collins
This is the typical unix way of doing things: you string together lots
of very specific tools to accomplish larger tasks. -- Vineet Kumar
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by "Jamin W. Collins" <jc...@asgardsrealm.net>.
On Wed, Jun 11, 2003 at 09:38:01PM -0500, cmpilato@collab.net wrote:
> cmpilato@collab.net writes:
>
> > Snatching now...
>
> Oh yeah, you're `changes' table is *way* horked. Geez, man.
So, is it "recoverable" horked, or "forget about it" horked?
I was just using svn_load_dirs. So, are we in agreement that this is
"corruption"?
--
Jamin W. Collins
Remember, root always has a loaded gun. Don't run around with it unless
you absolutely need it. -- Vineet Kumar
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by cm...@collab.net.
cmpilato@collab.net writes:
> cmpilato@collab.net writes:
>
> > "Jamin W. Collins" <jc...@asgardsrealm.net> writes:
> >
> > > > How big is the repos? Can I snatch it from a URL somewhere and take a
> > > > peek?
> > >
> > > It's almost 5 meg after tarring and gzipping. I've put it up on my
> > > personal d/l site:
> > >
> > > http://asgardsrealm.net/~jcollins/coding-horked-20030523.tar.gz
> >
> > Snatching now...
>
> Oh yeah, you're `changes' table is *way* horked. Geez, man.
...but I can now check out from your repository. :-D I'll send you
the repaired repos when I finish up some final testing.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: corrupted repository (was Re: the 0.24 release?)
Posted by Justin Erenkrantz <ju...@erenkrantz.com>.
--On Thursday, June 12, 2003 9:15 AM -0700 Michael Price
<mi...@computer.org> wrote:
> I've often wondered how the repo corruption was happening. I've used
> db 4.1 exclusively since the day it was released on multiple
> platforms using multiple ra protocols on each repo both at home and
> at work and haven't had a single corruption problem.
Just to chime in that I haven't had a single problem with BDB 4.1.25 on either
of my platforms (Solaris and Mac OS X). -- justin
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: corrupted repository (was Re: the 0.24 release?)
Posted by Michael Price <mi...@computer.org>.
--- Timothee Besset <tt...@idsoftware.com> wrote:
> I dunno if that's related, but I've seen similar things happen to
> my repository after 0.23 upgrade. It was because of db version
> mismatch. I used to have a 'default' build of httpd (which was
> probably using db 4 then), and svn 0.22 built against db 4.1
I've often wondered how the repo corruption was happening. I've used
db 4.1 exclusively since the day it was released on multiple
platforms using multiple ra protocols on each repo both at home and
at work and haven't had a single corruption problem.
Whenever I installed db 4.1 on a platform I first went and removed
all old db libraries and files. Any programs requiring db were then
recompiled.
Just a data point for people to consider.
Michael
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
corrupted repository (was Re: the 0.24 release?)
Posted by Timothee Besset <tt...@idsoftware.com>.
I dunno if that's related, but I've seen similar things happen to my
repository after 0.23 upgrade. It was because of db version mismatch. I
used to have a 'default' build of httpd (which was probably using db 4
then), and svn 0.22 built against db 4.1
When I switched to 0.23 with the same setup, things started to go bad. I
was not aware of the db 4 in httpd / db 4.1 in mod_dav_svn till then. I
rebuilt everything making sure httpd and svn where using db 4.1, and the
problems went away.
I am not sure this is related really. Just a hint. Anyway what I'm sure
of, is that svn doesn't check which version of db was used for apache, and
that can result in everything building fine, but heterogeneous db
versions.
TTimo
On 11 Jun 2003 22:27:01 -0500
cmpilato@collab.net wrote:
> cmpilato@collab.net writes:
>
> > cmpilato@collab.net writes:
> >
> > > "Jamin W. Collins" <jc...@asgardsrealm.net> writes:
> > >
> > > > > How big is the repos? Can I snatch it from a URL somewhere and take a
> > > > > peek?
> > > >
> > > > It's almost 5 meg after tarring and gzipping. I've put it up on my
> > > > personal d/l site:
> > > >
> > > > http://asgardsrealm.net/~jcollins/coding-horked-20030523.tar.gz
> > >
> > > Snatching now...
> >
> > Oh yeah, you're `changes' table is *way* horked. Geez, man.
>
> I'll be more specific. It appears that your `changes' table was
> somehow switched to a 'recno'-type database, whereas it is supposed to
> be a 'btree'. I don't know if this was the result of something gone
> horribly awry during the operation of Subversion, or if this was the
> result of some attempted recovery steps. Regardless, that file was
> junk.
>
> Fortunately, nothing in Subversion used that table except 'svn log',
> at least until I wrote the new svn_repos_replay(). So I replaced your
> `changes' table with one from a brand new repos, backdated my
> Subversion working copy until just before I made 'svnadmin dump' use
> the changes table, built, dumped your repos, and now I have a dumpfile
> reading for loading. It contains 26 revisions, which I trust sounds
> like the last known good state of your repository.
>
> I'll drop you a private mail with the dumpfile's location.
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
> For additional commands, e-mail: dev-help@subversion.tigris.org
>
>
--
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by Branko Čibej <br...@xbc.nu>.
Ben Collins-Sussman wrote:
>cmpilato@collab.net writes:
>
>
>
>>I'll be more specific. It appears that your `changes' table was
>>somehow switched to a 'recno'-type database, whereas it is supposed to
>>be a 'btree'.
>>
>>
>
>And this *exact* thing happened to another bdb-4.1 user.
>
>I think we're hot on the trail of something here.
>
>
Yes. In the meantime, we should tell people to stay away from bdb-4.1
(except on OS X, of course).
--
Brane Čibej <br...@xbc.nu> http://www.xbc.nu/brane/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by Ben Collins-Sussman <su...@collab.net>.
cmpilato@collab.net writes:
> I'll be more specific. It appears that your `changes' table was
> somehow switched to a 'recno'-type database, whereas it is supposed to
> be a 'btree'.
And this *exact* thing happened to another bdb-4.1 user.
I think we're hot on the trail of something here.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by cm...@collab.net.
cmpilato@collab.net writes:
> cmpilato@collab.net writes:
>
> > "Jamin W. Collins" <jc...@asgardsrealm.net> writes:
> >
> > > > How big is the repos? Can I snatch it from a URL somewhere and take a
> > > > peek?
> > >
> > > It's almost 5 meg after tarring and gzipping. I've put it up on my
> > > personal d/l site:
> > >
> > > http://asgardsrealm.net/~jcollins/coding-horked-20030523.tar.gz
> >
> > Snatching now...
>
> Oh yeah, you're `changes' table is *way* horked. Geez, man.
I'll be more specific. It appears that your `changes' table was
somehow switched to a 'recno'-type database, whereas it is supposed to
be a 'btree'. I don't know if this was the result of something gone
horribly awry during the operation of Subversion, or if this was the
result of some attempted recovery steps. Regardless, that file was
junk.
Fortunately, nothing in Subversion used that table except 'svn log',
at least until I wrote the new svn_repos_replay(). So I replaced your
`changes' table with one from a brand new repos, backdated my
Subversion working copy until just before I made 'svnadmin dump' use
the changes table, built, dumped your repos, and now I have a dumpfile
reading for loading. It contains 26 revisions, which I trust sounds
like the last known good state of your repository.
I'll drop you a private mail with the dumpfile's location.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by cm...@collab.net.
cmpilato@collab.net writes:
> "Jamin W. Collins" <jc...@asgardsrealm.net> writes:
>
> > > How big is the repos? Can I snatch it from a URL somewhere and take a
> > > peek?
> >
> > It's almost 5 meg after tarring and gzipping. I've put it up on my
> > personal d/l site:
> >
> > http://asgardsrealm.net/~jcollins/coding-horked-20030523.tar.gz
>
> Snatching now...
Oh yeah, you're `changes' table is *way* horked. Geez, man.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by cm...@collab.net.
"Jamin W. Collins" <jc...@asgardsrealm.net> writes:
> > How big is the repos? Can I snatch it from a URL somewhere and take a
> > peek?
>
> It's almost 5 meg after tarring and gzipping. I've put it up on my
> personal d/l site:
>
> http://asgardsrealm.net/~jcollins/coding-horked-20030523.tar.gz
Snatching now...
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by "Jamin W. Collins" <jc...@asgardsrealm.net>.
On Wed, Jun 11, 2003 at 09:17:05PM -0500, cmpilato@collab.net wrote:
> "Jamin W. Collins" <jc...@asgardsrealm.net> writes:
> > On Wed, Jun 11, 2003 at 08:49:37PM -0500, cmpilato@collab.net wrote:
> >
> > > Do you still have all the log.* files for the repos? Have you
> > > attempted a catastrophic recovery using those files?
> >
> > Yes, and yes. No luck. All forms of recovery that I'm aware of
> > (svnadmin and BDB) fail to restore the repository to a functional
> > state.
>
> How big is the repos? Can I snatch it from a URL somewhere and take a
> peek?
It's almost 5 meg after tarring and gzipping. I've put it up on my
personal d/l site:
http://asgardsrealm.net/~jcollins/coding-horked-20030523.tar.gz
--
Jamin W. Collins
Remember, root always has a loaded gun. Don't run around with it unless
you absolutely need it. -- Vineet Kumar
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by cm...@collab.net.
"Jamin W. Collins" <jc...@asgardsrealm.net> writes:
> On Wed, Jun 11, 2003 at 08:49:37PM -0500, cmpilato@collab.net wrote:
> > "Jamin W. Collins" <jc...@asgardsrealm.net> writes:
> >
> > > Would a dump of the database worked? Yea. However, for that to work, I
> > > would have to have a hook that dumped the repos after each and every
> > > successful commit, which I don't.
> >
> > Do you still have all the log.* files for the repos? Have you
> > attempted a catastrophic recovery using those files?
>
> Yes, and yes. No luck. All forms of recovery that I'm aware of
> (svnadmin and BDB) fail to restore the repository to a functional state.
How big is the repos? Can I snatch it from a URL somewhere and take a
peek?
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by "Jamin W. Collins" <jc...@asgardsrealm.net>.
On Wed, Jun 11, 2003 at 08:49:37PM -0500, cmpilato@collab.net wrote:
> "Jamin W. Collins" <jc...@asgardsrealm.net> writes:
>
> > Would a dump of the database worked? Yea. However, for that to work, I
> > would have to have a hook that dumped the repos after each and every
> > successful commit, which I don't.
>
> Do you still have all the log.* files for the repos? Have you
> attempted a catastrophic recovery using those files?
Yes, and yes. No luck. All forms of recovery that I'm aware of
(svnadmin and BDB) fail to restore the repository to a functional state.
--
Jamin W. Collins
This is the typical unix way of doing things: you string together lots
of very specific tools to accomplish larger tasks. -- Vineet Kumar
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by cm...@collab.net.
"Jamin W. Collins" <jc...@asgardsrealm.net> writes:
> Would a dump of the database worked? Yea. However, for that to work, I
> would have to have a hook that dumped the repos after each and every
> successful commit, which I don't.
Do you still have all the log.* files for the repos? Have you
attempted a catastrophic recovery using those files?
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by "Jamin W. Collins" <jc...@asgardsrealm.net>.
On Wed, Jun 11, 2003 at 02:35:21PM -0700, Greg Stein wrote:
> On Wed, Jun 11, 2003 at 02:09:29PM -0700, Blair Zajac wrote:
>
> > Where have you been :) There have been reports of svn/BDB 4.1.25
> > repositories getting corrupted probably once a month or so.
> >
> > Collab.net was going to put somebody onto tracking down this issue
> > but that hasn't happened yet.
>
> Eh? Is this the one that is generally typified by people using
> multiple access paths, and then you need to run recovery every now and
> then?
Nope, not in my case. Single access path, and more or less single user.
It's a personal system I have here in the house for myself and my wife.
We have a few repositories on it for different things. TMK, I was the
only one accessing the system at all that day.
> When I see the word "corruption", I think "data loss." Rather than
> "BDB is freaked out. Give it a 'lude."
There was data loss, complete and utter. The repos was (and still is
AFAIK) completely uncoverable. Luckily it was a repository that I was
using mainly to track changes another developer was making, and I was
able to retrieve periodic sources from him to recreate the repository
manually. So, I have the repository back, but not exactly as it was.
To me, that constitutes "data loss".
Would a dump of the database worked? Yea. However, for that to work, I
would have to have a hook that dumped the repos after each and every
successful commit, which I don't.
--
Jamin W. Collins
Remember, root always has a loaded gun. Don't run around with it unless
you absolutely need it. -- Vineet Kumar
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by Blair Zajac <bl...@orcaware.com>.
Greg Stein wrote:
>
> On Wed, Jun 11, 2003 at 02:09:29PM -0700, Blair Zajac wrote:
> > Greg Stein wrote:
> >...
> > > Hunh? What corruption are you referring to? I'm not aware of any problems at
> > > the moment. The only one that I know that has ever occurred was BDB 4.0 on
> > > the Mac OS X platform.
> >
> > Where have you been :) There have been reports of svn/BDB 4.1.25 repositories
> > getting corrupted probably once a month or so.
> >
> > Collab.net was going to put somebody onto tracking down this issue but that
> > hasn't happened yet.
>
> Eh? Is this the one that is generally typified by people using multiple
> access paths, and then you need to run recovery every now and then?
By access paths, you mean ra_dav and ra_svn. I only use ra_dav.
> When I see the word "corruption", I think "data loss." Rather than "BDB is
> freaked out. Give it a 'lude."
Most of my corruptions are such that "svnadmin recover" doesn't fix the
repository. I have to restore from a hot backup to get the repos running
again.
Best,
Blair
--
Blair Zajac <bl...@orcaware.com>
Plots of your system's performance - http://www.orcaware.com/orca/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by Greg Stein <gs...@lyra.org>.
On Wed, Jun 11, 2003 at 02:09:29PM -0700, Blair Zajac wrote:
> Greg Stein wrote:
>...
> > Hunh? What corruption are you referring to? I'm not aware of any problems at
> > the moment. The only one that I know that has ever occurred was BDB 4.0 on
> > the Mac OS X platform.
>
> Where have you been :) There have been reports of svn/BDB 4.1.25 repositories
> getting corrupted probably once a month or so.
>
> Collab.net was going to put somebody onto tracking down this issue but that
> hasn't happened yet.
Eh? Is this the one that is generally typified by people using multiple
access paths, and then you need to run recovery every now and then?
When I see the word "corruption", I think "data loss." Rather than "BDB is
freaked out. Give it a 'lude."
Cheers,
-g
--
Greg Stein, http://www.lyra.org/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by Branden Robinson <br...@deadbeast.net>.
On Thu, Jun 12, 2003 at 02:22:59PM -0400, Scott Moynes wrote:
> We recently had database corruption that was not recoverable without
> recreating the repository and loading from a dump. We noticed svn
> would hang trying to open the database, and looking through debian's
> bug database we found this:
>
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=185033
Hmm, a bug that was fixed by a patch from Sleepycat that was applied to
the 4.1.25-3 package, which was released in March.
I use BDB 4.1 on Debian and have never yet seen this problem (knock
wood) with my repo. I did not create the repository until April.
Access to my repo is via svn+ssh and local methods only.
("My repo" -> http://deadbeast.net/cgi-bin/viewcvs.cgi/)
--
G. Branden Robinson | The basic test of freedom is
Free Software Developer | perhaps less in what we are free to
branden@deadbeast.net | do than in what we are free not to
http://deadbeast.net/~branden/ | do. -- Eric Hoffer
Re: the 0.24 release?
Posted by Michael Price <mi...@computer.org>.
--- Scott Moynes <sm...@nexus.carleton.ca> wrote:
> We recently had database corruption that was not recoverable
> without recreating the repository and loading from a dump. We
> noticed svn would hang trying to open the database, and looking
> through debian's bug database we found this:
>
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=185033
>
> It is too early to know if this was in fact the cause of our
> corruption or not, though.
To summarize:
http://ftp.debian.org/debian/pool/main/d/db4.1/db4.1_4.1.25-4.diff.gz
contains only one db source file diff (which is attached). This diff
is a combination of patch 1 from Sleepycat
(http://www.sleepycat.com/update/4.1.25/patch.4.1.25.html) plus
another bit (apparently from Sleepycat directly to the debian people)
that fixes the abort issue. I don't know why its not also listed on
Sleepycat's patch page.
Michael
Re: the 0.24 release?
Posted by Scott Moynes <sm...@nexus.carleton.ca>.
* Blair Zajac (blair@orcaware.com) wrote:
> Where have you been :) There have been reports of svn/BDB 4.1.25
> repositories getting corrupted probably once a month or so.
>
> Collab.net was going to put somebody onto tracking down this issue
> but that hasn't happened yet.
We recently had database corruption that was not recoverable without
recreating the repository and loading from a dump. We noticed svn
would hang trying to open the database, and looking through debian's
bug database we found this:
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=185033
It is too early to know if this was in fact the cause of our
corruption or not, though.
--
Scott Moynes http://www.icculus.org/openbox/
"Computer science is as much about computers
as astronomy is about telescopes." -- Dijkstra
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by Blair Zajac <bl...@orcaware.com>.
Greg Stein wrote:
>
> On Wed, Jun 11, 2003 at 01:54:06PM -0600, Jamin W. Collins wrote:
> > On Wed, Jun 11, 2003 at 12:47:37PM -0700, Greg Stein wrote:
> >
> > > I'm not aware of any major issue that needs to be fixed for the
> > > release, so it can basically be cut any time now.
> >
> > Just a user's point of view, but what about the Berkeley DB corruption?
> > Is that still a specter or has that been located and resolved?
>
> Hunh? What corruption are you referring to? I'm not aware of any problems at
> the moment. The only one that I know that has ever occurred was BDB 4.0 on
> the Mac OS X platform.
Where have you been :) There have been reports of svn/BDB 4.1.25 repositories
getting corrupted probably once a month or so.
Collab.net was going to put somebody onto tracking down this issue but that
hasn't happened yet.
Best,
Blair
--
Blair Zajac <bl...@orcaware.com>
Plots of your system's performance - http://www.orcaware.com/orca/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by Greg Stein <gs...@lyra.org>.
On Wed, Jun 11, 2003 at 01:54:06PM -0600, Jamin W. Collins wrote:
> On Wed, Jun 11, 2003 at 12:47:37PM -0700, Greg Stein wrote:
>
> > I'm not aware of any major issue that needs to be fixed for the
> > release, so it can basically be cut any time now.
>
> Just a user's point of view, but what about the Berkeley DB corruption?
> Is that still a specter or has that been located and resolved?
Hunh? What corruption are you referring to? I'm not aware of any problems at
the moment. The only one that I know that has ever occurred was BDB 4.0 on
the Mac OS X platform.
Cheers,
-g
--
Greg Stein, http://www.lyra.org/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by "Jamin W. Collins" <jc...@asgardsrealm.net>.
On Wed, Jun 11, 2003 at 12:47:37PM -0700, Greg Stein wrote:
> I'm not aware of any major issue that needs to be fixed for the
> release, so it can basically be cut any time now.
Just a user's point of view, but what about the Berkeley DB corruption?
Is that still a specter or has that been located and resolved?
--
Jamin W. Collins
Remember, root always has a loaded gun. Don't run around with it unless
you absolutely need it. -- Vineet Kumar
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by Greg Hudson <gh...@MIT.EDU>.
On Wed, 2003-06-11 at 16:01, Ben Collins-Sussman wrote:
> But I wonder if it makes sense to let kfogel finish his branch/tag
> support in cvs2svn.py? He's been working hard on that, and seems to
> be a day or two away from finishing. Is it worth delaying the 0.24 on
> that task? It would be another huge feather in the 0.24 cap.
Owing to the complexity of the problem, I'd say if the code isn't 100%
done and tested now, there's a substantial probability that it won't be
done for another week or two. And our ideal 0.x release cycle is only a
couple of weeks long. So, I'd say it doesn't make sense to delay.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by Philip Martin <ph...@codematters.co.uk>.
Ben Collins-Sussman <su...@collab.net> writes:
> I wanted to point out that there's been a whole lot of great
> development, even if most of the 0.24 "scheduled" issues are still
> open:
>
> - lots of (incompatible) improvements to svnserve's protocol
> - a whole new path-based authorization module for apache
> - a new svn_repos_replay() routine, which makes dumping and svnlook faster
> - 'svn switch --relocate'
> - buncha bugfixes too
One other thing that may not be immediately obvious from the 0.23 to
HEAD changelog is the progress that has been made towards solving
issue 773--although the scaling problem hasn't been solved, memory use
has been substantially reduced.
--
Philip Martin
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by kf...@collab.net.
Roman Neuhauser <ne...@bellavista.cz> writes:
> what's the current estimate? this problem is the only thing that
> keeps us from switching.
More than a day, less than a month.
Sorry can't help you more; the problem involves a lot of edge cases.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by Roman Neuhauser <ne...@bellavista.cz>.
# kfogel@collab.net / 2003-06-12 09:58:23 -0500:
> Jani Averbach <ja...@cc.jyu.fi> writes:
> > Because if 0.24 will be shipped with cvs2svn.py that is capable to do
> > branch and tags it _would_ be a bite for people to try svn by converting
> > cvs-repos to svn. So if our next release will be fs-schema incompatible,
> > is it better to release 0.24 as it is now, and cvs2svn.py and fs-schema
> > together, isn't?
>
> cvs2svn will not be capabable of branches/tags in 0.24.
what's the current estimate? this problem is the only thing that
keeps us from switching.
--
If you cc me or remove the list(s) completely I'll most likely ignore
your message. see http://www.eyrie.org./~eagle/faqs/questions.html
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by kf...@collab.net.
Jani Averbach <ja...@cc.jyu.fi> writes:
> Because if 0.24 will be shipped with cvs2svn.py that is capable to do
> branch and tags it _would_ be a bite for people to try svn by converting
> cvs-repos to svn. So if our next release will be fs-schema incompatible,
> is it better to release 0.24 as it is now, and cvs2svn.py and fs-schema
> together, isn't?
cvs2svn will not be capabable of branches/tags in 0.24.
-Karl
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by cm...@collab.net.
Jani Averbach <ja...@cc.jyu.fi> writes:
> Hmm, will there be a second, planned fs-schema change in near or not so
> near future?
I expect no futher schema changes after this one, at least not until
post-1.0. But I'm not signing any legal statement to that effect. :-)
> Because if 0.24 will be shipped with cvs2svn.py that is capable to do
> branch and tags it _would_ be a bite for people to try svn by converting
> cvs-repos to svn. So if our next release will be fs-schema incompatible,
> is it better to release 0.24 as it is now, and cvs2svn.py and fs-schema
> together, isn't?
That's an interesting thing to consider. I don't think these need to
be necessarily tied -- it should be enough to note in the cvs2svn
documentation and the release notes that folks might want to wait
until the schema changes are merged into the main trunk before
converting their cvs repositories over.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by Jani Averbach <ja...@cc.jyu.fi>.
On 11 Jun 2003, Ben Collins-Sussman wrote:
> But I wonder if it makes sense to let kfogel finish his branch/tag
> support in cvs2svn.py? He's been working hard on that, and seems to
> be a day or two away from finishing. Is it worth delaying the 0.24 on
> that task? It would be another huge feather in the 0.24 cap.
Hmm, will there be a second, planned fs-schema change in near or not so
near future?
Because if 0.24 will be shipped with cvs2svn.py that is capable to do
branch and tags it _would_ be a bite for people to try svn by converting
cvs-repos to svn. So if our next release will be fs-schema incompatible,
is it better to release 0.24 as it is now, and cvs2svn.py and fs-schema
together, isn't?
However this is just a thought.
My own bite to use svn was Marko Macek's "branch and tags"-patch and it
was the reason to make the switch.
BR, Jani
--
Jani Averbach
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by Ben Collins-Sussman <su...@collab.net>.
Greg Stein <gs...@lyra.org> writes:
> The 0.24 release is scheduled for tomorrow. While we didn't happen to knock
> out the items scheduled in IZ for the release, it still seems very important
> to *have* one because of all the work that has gone into it. Ben points out
> that we actually forked for the 0.23 release about six weeks ago.
I wanted to point out that there's been a whole lot of great
development, even if most of the 0.24 "scheduled" issues are still
open:
- lots of (incompatible) improvements to svnserve's protocol
- a whole new path-based authorization module for apache
- a new svn_repos_replay() routine, which makes dumping and svnlook faster
- 'svn switch --relocate'
- buncha bugfixes too
I agree with gstein, that we should merge cmpilato's big fs-schema
changes *after* the 0.24 release, to let them "settle in" for a bit
with developers, rather than unleashing them on users immediately.
But I wonder if it makes sense to let kfogel finish his branch/tag
support in cvs2svn.py? He's been working hard on that, and seems to
be a day or two away from finishing. Is it worth delaying the 0.24 on
that task? It would be another huge feather in the 0.24 cap.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by Branko Čibej <br...@xbc.nu>.
Greg Stein wrote:
>Michael,
>
>The 0.24 release is scheduled for tomorrow. While we didn't happen to knock
>out the items scheduled in IZ for the release, it still seems very important
>to *have* one because of all the work that has gone into it. Ben points out
>that we actually forked for the 0.23 release about six weeks ago.
>
>And while Mike has got his FS work nearly ready to merge into the trunk, it
>would seem appropriate to let that bake in the trunk for developers rather
>than having our users be the first people to try it out. Thus, releasing
>0.24 before that merge would seem to be appropriate.
>
>I'm not aware of any major issue that needs to be fixed for the release, so
>it can basically be cut any time now.
>
>Are you available to do the release tonite / tomorrow?
>
>
BTW, before anyone starts to roll 0.24, I'd like to get some build
changes in so that mod_authz_svn.so gets built on Windows. The changes
are already in my tree, I just have to do some testing before commit --
this evening, CEST.
--
Brane Čibej <br...@xbc.nu> http://www.xbc.nu/brane/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: the 0.24 release?
Posted by Michael Price <mi...@computer.org>.
> I'm not aware of any major issue that needs to be fixed for the
> release, so it can basically be cut any time now.
>
> Are you available to do the release tonite / tomorrow?
I can update the CHANGES file tonight but I won't be available to
roll a release until tomorrow night. Wednesday nights are busy for
me.
Michael
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org