You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Christian Daudt <cs...@daudt.org> on 2003/02/27 19:03:57 UTC

timeout problem

Hi,
 I've just started trying out subversion (looks great btw) and I've run across 
a problem (which seems to be a known problem but I'll throw the info anyways 
so people that want to look into it have more data). Here is what I did on a 
freshly created repository:
 svn import <import-point> <linux-kernel-2.5.0>
 svn copy <import-point> <2.5.0-tagpoint>
 svn_load_dirs.pl <import-point> <2.5.10-tree>
 at this point I get a failure (45 minutes into the process):

 subversion/libsvm_client/commit.c:670: (apr_err=175002)
 svn: RA layer request failed
 svn: Commit failed (details follow):
 subversion/libsvn_ra_dav/util.c:393 (apr_err=175002)
 svn: MERGE request failed on /repository/linux/import/v2.5/current
 subversion/libsvn_ra_dav/util.c:81: (apr_err=175002)
 svn: MERGE of /repository/linux/import/v2.5/current: timed out waiting for 
server
 Press return to quit and clean up svn working directory:

 I've attached my httpd.conf because I had to tweak it into not being such a 
memory hog (my server has 512MB of RAM and I was running 800MB into swap with 
the default config). I'm running the latest svn from the repository 
(Revision: 5125)

 thanks,
   csd

Re: timeout problem

Posted by Christian Daudt <cs...@daudt.org>.
 Just finished re-running with r5153 and it's gotten past the point of error 
with the local test. Thanks !
  I'll pound it some more locally and then re-run all the tests via http and 
provide results later (probably tomorrow because each one of these runs takes 
a good couple of hours...).

 thanks again,
   csd


On Thursday 27 February 2003 22:26, cmpilato@collab.net wrote:
> cmpilato@collab.net writes:
> > Christian Daudt <cs...@daudt.org> writes:
> > > Ok,
> > >  ran into a completely different error doing the same test locally:
> > >
> > > subversion/libsvn_client/commit.c:670: (apr_err=160029)
> > > svn: Berkeley DB error
> > > svn: Commit failed (details follow):
> > > subversion/libsvn_fs/tree.c:2364: (apr_err=160029)
> > > svn: Commit succeeded, deltification failed
> > > subversion/libsvn_fs/bdb/bdb-err.c:61: (apr_err=160029)
> > > svn: Berkeley DB error while deleting string for filesystem
> > > /home/csd/tmp/repo/db:
> > > Cannot allocate memory
> > >
> > >
> > > Note that it didn't look like it had really run out of memory but to
> > > be sure I added 4GB to my swap space and reran the test and I still
> > > got the same result. It happened at the same point that the first
> > > one did (i.e. doing the merge of the 2nd imported tree)
> >
> > If you are able to do so, I'd love to hear if upgrading to Subversion
> > r5147 or later solves this probelm for you.
>
> Er.  Make that r5148.
>
> {mumble, grumble, brain-o, rumble, tumble...}
>
> ---------------------------------------------------------------------
> 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: timeout problem

Posted by "Glenn A. Thompson" <gt...@cdr.net>.
> 
>
>Er.  Make that r5148.
>
* gat puts CMike in infinite loop so he can catch up:-)

Uh, you boys been busy.

>  
>


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: timeout problem

Posted by cm...@collab.net.
cmpilato@collab.net writes:

> Christian Daudt <cs...@daudt.org> writes:
> 
> > Ok,
> >  ran into a completely different error doing the same test locally:
> > 
> > subversion/libsvn_client/commit.c:670: (apr_err=160029)
> > svn: Berkeley DB error
> > svn: Commit failed (details follow):
> > subversion/libsvn_fs/tree.c:2364: (apr_err=160029)
> > svn: Commit succeeded, deltification failed
> > subversion/libsvn_fs/bdb/bdb-err.c:61: (apr_err=160029)
> > svn: Berkeley DB error while deleting string for filesystem 
> > /home/csd/tmp/repo/db:
> > Cannot allocate memory
> > 
> > 
> > Note that it didn't look like it had really run out of memory but to
> > be sure I added 4GB to my swap space and reran the test and I still
> > got the same result. It happened at the same point that the first
> > one did (i.e. doing the merge of the 2nd imported tree)
> 
> If you are able to do so, I'd love to hear if upgrading to Subversion
> r5147 or later solves this probelm for you.

Er.  Make that r5148.

{mumble, grumble, brain-o, rumble, tumble...}

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: timeout problem

Posted by cm...@collab.net.
Christian Daudt <cs...@daudt.org> writes:

> Ok,
>  ran into a completely different error doing the same test locally:
> 
> subversion/libsvn_client/commit.c:670: (apr_err=160029)
> svn: Berkeley DB error
> svn: Commit failed (details follow):
> subversion/libsvn_fs/tree.c:2364: (apr_err=160029)
> svn: Commit succeeded, deltification failed
> subversion/libsvn_fs/bdb/bdb-err.c:61: (apr_err=160029)
> svn: Berkeley DB error while deleting string for filesystem 
> /home/csd/tmp/repo/db:
> Cannot allocate memory
> 
> 
> Note that it didn't look like it had really run out of memory but to
> be sure I added 4GB to my swap space and reran the test and I still
> got the same result. It happened at the same point that the first
> one did (i.e. doing the merge of the 2nd imported tree)

If you are able to do so, I'd love to hear if upgrading to Subversion
r5147 or later solves this probelm for you.

Here's what I did:

   Stop performing deltification as part of one big whoppin' Berkeley
   transaction.
   
   * subversion/libsvn_fs/tree.c
     (struct txn_deltify_args, txn_body_txn_deltify): New.
     (deltify_mutable): Was deltify_if_mutable_under_txn_id(), and hack
       this thing up so that it no longer uses a single trail for the
       whole deltification process.
     (txn_body_deltify_committed): Remove.
     (svn_fs_commit_txn): Use the new and improved deltify_mutable().

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: timeout problem

Posted by cm...@collab.net.
Christian Daudt <cs...@daudt.org> writes:

> Ok,
>  ran into a completely different error doing the same test locally:
> 
> subversion/libsvn_client/commit.c:670: (apr_err=160029)
> svn: Berkeley DB error
> svn: Commit failed (details follow):
> subversion/libsvn_fs/tree.c:2364: (apr_err=160029)
> svn: Commit succeeded, deltification failed
> subversion/libsvn_fs/bdb/bdb-err.c:61: (apr_err=160029)
> svn: Berkeley DB error while deleting string for filesystem 
> /home/csd/tmp/repo/db:
> Cannot allocate memory

Almost every time I've seen this, it's been about database locking.
In this case, I can almost *guarantee* that you saw the problem
because deltification happens inside one big fat Berkeley transaction,
and you just flat ran out of locks.

I, uh... should probably just fix that tonight, shouldn't I?


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: timeout problem

Posted by Christian Daudt <cs...@daudt.org>.
Ok,
 ran into a completely different error doing the same test locally:

subversion/libsvn_client/commit.c:670: (apr_err=160029)
svn: Berkeley DB error
svn: Commit failed (details follow):
subversion/libsvn_fs/tree.c:2364: (apr_err=160029)
svn: Commit succeeded, deltification failed
subversion/libsvn_fs/bdb/bdb-err.c:61: (apr_err=160029)
svn: Berkeley DB error while deleting string for filesystem 
/home/csd/tmp/repo/db:
Cannot allocate memory


Note that it didn't look like it had really run out of memory but to be sure I 
added 4GB to my swap space and reran the test and I still got the same 
result. It happened at the same point that the first one did (i.e. doing the 
merge of the 2nd imported tree)

 On a side note, I've started looking at subversion because it seems like a 
nice fit to maintain the kernel development we're doing here. I was wondering 
if there are others out there also maintaining/planning to maintain linux 
kernels with subversion. I saw the discussion on Linus switching to it or 
not, but there are other people doing development with very different needs 
from what Linus does.

 cheers,
   csd

On Thursday 27 February 2003 13:06, Christian Daudt wrote:
> It's not 1077 because nothing ran out of memory when I got this error (it
> was after I reconfigured httpd to not be such a hog). More likely 995/773.
> I'll try rerunning it locally and see what I get.
>
>  csd
>
> On Thursday 27 February 2003 11:50, Karl Fogel wrote:
> > Christian Daudt <cs...@daudt.org> writes:
> > >  svn import <import-point> <linux-kernel-2.5.0>
> > >  svn copy <import-point> <2.5.0-tagpoint>
> > >  svn_load_dirs.pl <import-point> <2.5.10-tree>
> > >  at this point I get a failure (45 minutes into the process):
> > >
> > >  subversion/libsvm_client/commit.c:670: (apr_err=175002)
> > >  svn: RA layer request failed
> > >  svn: Commit failed (details follow):
> > >  subversion/libsvn_ra_dav/util.c:393 (apr_err=175002)
> > >  svn: MERGE request failed on /repository/linux/import/v2.5/current
> > >  subversion/libsvn_ra_dav/util.c:81: (apr_err=175002)
> > >  svn: MERGE of /repository/linux/import/v2.5/current: timed out waiting
> > > for server
> > >  Press return to quit and clean up svn working directory:
> >
> > Could this be one of
> >
> >    http://subversion.tigris.org/issues/show_bug.cgi?id=1077
> >    http://subversion.tigris.org/issues/show_bug.cgi?id=995
> >    http://subversion.tigris.org/issues/show_bug.cgi?id=773
> >
> > ?  (The first seems most likely to me.)
> >
> > -Karl
> >
> > ---------------------------------------------------------------------
> > 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


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: timeout problem

Posted by Christian Daudt <cs...@daudt.org>.
It's not 1077 because nothing ran out of memory when I got this error (it was 
after I reconfigured httpd to not be such a hog). More likely 995/773. I'll 
try rerunning it locally and see what I get.

 csd

On Thursday 27 February 2003 11:50, Karl Fogel wrote:
> Christian Daudt <cs...@daudt.org> writes:
> >  svn import <import-point> <linux-kernel-2.5.0>
> >  svn copy <import-point> <2.5.0-tagpoint>
> >  svn_load_dirs.pl <import-point> <2.5.10-tree>
> >  at this point I get a failure (45 minutes into the process):
> >
> >  subversion/libsvm_client/commit.c:670: (apr_err=175002)
> >  svn: RA layer request failed
> >  svn: Commit failed (details follow):
> >  subversion/libsvn_ra_dav/util.c:393 (apr_err=175002)
> >  svn: MERGE request failed on /repository/linux/import/v2.5/current
> >  subversion/libsvn_ra_dav/util.c:81: (apr_err=175002)
> >  svn: MERGE of /repository/linux/import/v2.5/current: timed out waiting
> > for server
> >  Press return to quit and clean up svn working directory:
>
> Could this be one of
>
>    http://subversion.tigris.org/issues/show_bug.cgi?id=1077
>    http://subversion.tigris.org/issues/show_bug.cgi?id=995
>    http://subversion.tigris.org/issues/show_bug.cgi?id=773
>
> ?  (The first seems most likely to me.)
>
> -Karl
>
> ---------------------------------------------------------------------
> 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: timeout problem

Posted by Karl Fogel <kf...@newton.ch.collab.net>.
Christian Daudt <cs...@daudt.org> writes:
>  svn import <import-point> <linux-kernel-2.5.0>
>  svn copy <import-point> <2.5.0-tagpoint>
>  svn_load_dirs.pl <import-point> <2.5.10-tree>
>  at this point I get a failure (45 minutes into the process):
> 
>  subversion/libsvm_client/commit.c:670: (apr_err=175002)
>  svn: RA layer request failed
>  svn: Commit failed (details follow):
>  subversion/libsvn_ra_dav/util.c:393 (apr_err=175002)
>  svn: MERGE request failed on /repository/linux/import/v2.5/current
>  subversion/libsvn_ra_dav/util.c:81: (apr_err=175002)
>  svn: MERGE of /repository/linux/import/v2.5/current: timed out waiting for 
> server
>  Press return to quit and clean up svn working directory:

Could this be one of

   http://subversion.tigris.org/issues/show_bug.cgi?id=1077
   http://subversion.tigris.org/issues/show_bug.cgi?id=995
   http://subversion.tigris.org/issues/show_bug.cgi?id=773

?  (The first seems most likely to me.)

-Karl

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org