You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by da...@fastmail.co.uk on 2008/07/31 13:23:22 UTC

svn log hangs munching cpu

Hi,

following is top output:

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
31151 plgtest   25   0  7316 2564 2024 R   64  0.1   1215:37 svn
 6416 plgtest   25   0  7316 2564 2024 R   55  0.1   1489:08 svn
24938 plgtest   25   0  7316 2564 2024 R   50  0.1   1538:18 svn
32493 plgtest   25   0  7312 2556 2024 R   49  0.1 746:08.08 svn
  800 plgtest   25   0  7316 2564 2024 R   47  0.1 747:46.63 svn
 5718 plgtest   25   0  7312 2560 2024 R   46  0.1   1318:43 svn
 2382 plgtest   25   0  7312 2560 2024 R   45  0.1   1268:37 svn

the command lines for the hung svn processes are of the form:

svn log --incremental http://... -r$REV

and, usually (even on the same box), running the exact same command
is fast (0.08 seconds), so this seems not reproducable. It may have
been triggered by network problems, but there's no way to know now.

Is this a known problem? Should I raise an issue?

client is svn v1.4.2 running on Debian, server is v1.4.3 via Apache.

thanks

Dave
-- 
  
  dave_rodgman@fastmail.co.uk


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

Re: svn log hangs munching cpu

Posted by da...@fastmail.co.uk.
On Fri, 1 Aug 2008 19:32:54 +0200, "Erik Huelsmann" <eh...@gmail.com>
said:
> On 7/31/08, dave_rodgman@fastmail.co.uk <da...@fastmail.co.uk>
> wrote:
> > Hi,
> >
> > following is top output:
> >
> >  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
> > 31151 plgtest   25   0  7316 2564 2024 R   64  0.1   1215:37 svn
> >  6416 plgtest   25   0  7316 2564 2024 R   55  0.1   1489:08 svn
> > 24938 plgtest   25   0  7316 2564 2024 R   50  0.1   1538:18 svn
> > 32493 plgtest   25   0  7312 2556 2024 R   49  0.1 746:08.08 svn
> >  800 plgtest   25   0  7316 2564 2024 R   47  0.1 747:46.63 svn
> >  5718 plgtest   25   0  7312 2560 2024 R   46  0.1   1318:43 svn
> >  2382 plgtest   25   0  7312 2560 2024 R   45  0.1   1268:37 svn
> >
> > the command lines for the hung svn processes are of the form:
> >
> > svn log --incremental http://... -r$REV
> >
> > and, usually (even on the same box), running the exact same command
> 
> exactly the same command, or using a different protocol? (ie svn log
> --incremental file://...)?

Hi Erik,

Yes, exactly the same, via the http protocol. I just did
"cat /proc/800/cmdline" and typed that into the command line.

I'm aware that authorisation can be expensive, but I don't think it's
the problem in this case.

- normally an identical command succeeds very quickly
- the hung commands have been running for 746 hours in some cases

so, I think there is some other issue here.

Dave

> > is fast (0.08 seconds), so this seems not reproducable. It may have
> > been triggered by network problems, but there's no way to know now.
> >
> > Is this a known problem? Should I raise an issue?
> 
> Yes, it's a known issue. Please don't raise one.
> 
> > client is svn v1.4.2 running on Debian, server is v1.4.3 via Apache.
> 
> I'm sorry to have to tell you, but if you mean that it's *not* exactly
> the same command (ie using different protocols) you should *really*
> have contacted the users@ list: the problem you're reporting is
> addressed here:
> http://svnbook.red-bean.com/nightly/en/svn.serverconfig.httpd.html#svn.serverconfig.httpd.authz.pathauthzoff
> 
> Unfortunately, there's not much we can do about this issue, unless you
> disable authorization.
> 
> HTH,
> 
> Erik.
-- 
  
  dave_rodgman@fastmail.co.uk


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

Re: svn log hangs munching cpu

Posted by Erik Huelsmann <eh...@gmail.com>.
On 7/31/08, dave_rodgman@fastmail.co.uk <da...@fastmail.co.uk> wrote:
> Hi,
>
> following is top output:
>
>  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
> 31151 plgtest   25   0  7316 2564 2024 R   64  0.1   1215:37 svn
>  6416 plgtest   25   0  7316 2564 2024 R   55  0.1   1489:08 svn
> 24938 plgtest   25   0  7316 2564 2024 R   50  0.1   1538:18 svn
> 32493 plgtest   25   0  7312 2556 2024 R   49  0.1 746:08.08 svn
>  800 plgtest   25   0  7316 2564 2024 R   47  0.1 747:46.63 svn
>  5718 plgtest   25   0  7312 2560 2024 R   46  0.1   1318:43 svn
>  2382 plgtest   25   0  7312 2560 2024 R   45  0.1   1268:37 svn
>
> the command lines for the hung svn processes are of the form:
>
> svn log --incremental http://... -r$REV
>
> and, usually (even on the same box), running the exact same command

exactly the same command, or using a different protocol? (ie svn log
--incremental file://...)?

> is fast (0.08 seconds), so this seems not reproducable. It may have
> been triggered by network problems, but there's no way to know now.
>
> Is this a known problem? Should I raise an issue?

Yes, it's a known issue. Please don't raise one.

> client is svn v1.4.2 running on Debian, server is v1.4.3 via Apache.

I'm sorry to have to tell you, but if you mean that it's *not* exactly
the same command (ie using different protocols) you should *really*
have contacted the users@ list: the problem you're reporting is
addressed here:
http://svnbook.red-bean.com/nightly/en/svn.serverconfig.httpd.html#svn.serverconfig.httpd.authz.pathauthzoff

Unfortunately, there's not much we can do about this issue, unless you
disable authorization.

HTH,

Erik.

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