You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Sean Russell <se...@germane-software.com> on 2002/05/28 17:54:10 UTC

Server hangs (r2019) (possibly re: possible database corruption; need advice)

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

I've been having the subversion server hang on me quite a bit lately, and 
requiring  a run of db_restore.  I think this is the same issue as cmpilato 
has been working on lately (msg ID 74464), but r2019 doesn't seem to fix the 
problem.

The problem seems to appear for me after/during 'svn cp <dav_url> <dav_url>'.  
I'm not seeing any trace information on the server; it juts silently hangs, 
and further access attempts of any type to the server all hang -- even ones 
via ra_local.  Restarting Apache doesn't seem to help.  I must run db_recover 
on the database and restart Apache, and then everything works fine for a 
while.

It doesn't happen after every 'svn cp', and it doesn't require mvs or dels; 
just 'svn cp' on directories, separated by some 'ci's and possibly 'add's.  
After an arbitrary number of these copies, I get something like this:

svn_error: #21093 : <RA layer request failed>
  MERGE request failed on /svn/repos/rexml

svn_error: #21093 : <RA layer request failed>
  MERGE of /svn/repos/rexml: timed out waiting for server

I /think/ you can reproduce this by trying a copy and breaking in the middle 
of the copy, or trying a copy that will fail (such as copying to a URL that 
already exists).  If this doesn't cause an error, then I only know that the 
error occurs after 'n' copies.  I haven't had it hang on a co or ci, yet.

This happens to me fairly regularly, so if there's some other trace 
information I can provide when it does happen, please let me know.  I don't 
know where the "stack analizer scripts" are located.  How much use would the 
pstack output be?  strace -p <pid>? 

Here's the db_recover output, for what good it is.

[root@maus subversion]# db_recover -h /home/subversion/db -ve
db_recover: Finding last valid log LSN: file: 9 offset 3595927
db_recover: Checkpoint at: [9][3595466]
db_recover: Checkpoint LSN: [9][3504542]
db_recover: Previous checkpoint: [9][3502579]
db_recover: Checkpoint at: [9][3502579]
db_recover: Checkpoint LSN: [9][3502579]
db_recover: Previous checkpoint: [9][3492852]
db_recover: Recovery starting from [9][3502579]
db_recover: Recovery complete at Tue May 28 09:54:15 2002
db_recover: Maximum transaction ID 8002c173 Recovery checkpoint [9][3596388]
db_recover: Recovery complete at Tue May 28 09:54:15 2002
db_recover: Maximum transaction id 80000000 Recovery checkpoint [9][3596388]

- --- SER
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.7 (GNU/Linux)

iD8DBQE888RCP0KxygnleI8RAhFpAJ9KUAA1+ooZ2Ot2nN/cEhjbbODKnQCfZeYF
W0mb8EWKVlvWYf0otBgijy4=
=VJTu
-----END PGP SIGNATURE-----

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

Re: Server hangs (r2019) (possibly re: possible database corruption; need advice)

Posted by Ben Collins-Sussman <su...@collab.net>.
Sean Russell <se...@germane-software.com> writes:

> Je Mar Majo 28 2002 11:18, cmpilato@collab.net skribis:
> > Um...sorry to ask a silly question, but: did you actually port the
> > change in r2019 to the "trunk" subversion code?  I committed that fix
> > against the branch I'm currently working on.
> 
> Not a silly question at all.  No, I didn't.  :-/  I thought you had done that.  
> I think I made the mistake of thinking that the trunk was at the most recent 
> revision -- I saw the 2019 and assumed that was what I had.
> 
> I guess I don't have the fix, so ignore my posting.  Sorry for the noise.

Update your working copy, I just ported the fix from cmpilato's branch
to /trunk.  (r2021)


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

Re: Server hangs (r2019) (possibly re: possible database corruption; need advice)

Posted by Sean Russell <se...@germane-software.com>.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Je Mar Majo 28 2002 11:18, cmpilato@collab.net skribis:
> Um...sorry to ask a silly question, but: did you actually port the
> change in r2019 to the "trunk" subversion code?  I committed that fix
> against the branch I'm currently working on.

Not a silly question at all.  No, I didn't.  :-/  I thought you had done that.  
I think I made the mistake of thinking that the trunk was at the most recent 
revision -- I saw the 2019 and assumed that was what I had.

I guess I don't have the fix, so ignore my posting.  Sorry for the noise.

- -- 
 |..  War is God's way of teaching Americans geography.
<|>   --- Ambrose Bierce
/|\   
/|    
 |         
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.7 (GNU/Linux)

iD8DBQE889UgP0KxygnleI8RAgjuAJsFjX3cdxUlAwC/uE8qikUc6ghVGQCgkKJn
8lqTOIYnLTOpbLUgMm9MNO4=
=yrpM
-----END PGP SIGNATURE-----

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

Re: Server hangs (r2019) (possibly re: possible database corruption; need advice)

Posted by cm...@collab.net.
Sean Russell <se...@germane-software.com> writes:

> I've been having the subversion server hang on me quite a bit lately, and 
> requiring  a run of db_restore.  I think this is the same issue as cmpilato 
> has been working on lately (msg ID 74464), but r2019 doesn't seem to fix the 
> problem.

Um...sorry to ask a silly question, but: did you actually port the
change in r2019 to the "trunk" subversion code?  I committed that fix
against the branch I'm currently working on.

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