You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Nick Thompson <ni...@lsi.com> on 2007/09/24 08:47:45 UTC
pre-commit stderr output sometimes lost
Subversion 1.4.3 on RHEL3 with Apache 2.0.55
Hi,
Just occasionally, we see that the error output from pre-commit gets
lost. The error is flagged to the client which sees a 409 error
message, but non of the really useful output that the hook generates.
The error text is logged in the apache error_log though:
[Mon Sep 24 09:25:17 2007] [error] [client xxx.xx.xx.xxx] 'pre-commit'
hook failed with error output:\nDetected reference to blah blah in
file.x at line 2\nPlease replace with something else\n [409,
#165001]
The user at the client has no idea what to do next until we send them
the error_log message from apache.
I can't determine what the trigger for this behavior is, but it
doesn't always happen. Is it Subversion or Apache problem? I can
upgrade either or both if needs be.
Thanks,
--
> Nick Thompson
> LSI Technology Ltd
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: pre-commit stderr output sometimes lost
Posted by Erik Huelsmann <eh...@gmail.com>.
On 9/24/07, Dave Rodgman <da...@fastmail.co.uk> wrote:
> Erik Huelsmann wrote:
> > On 9/24/07, Dave Rodgman <da...@fastmail.co.uk> wrote:
> >
> >> Nick Thompson wrote:
> >>
> >>> Subversion 1.4.3 on RHEL3 with Apache 2.0.55
> >>>
> >>> Hi,
> >>>
> >>> Just occasionally, we see that the error output from pre-commit gets
> >>> lost. The error is flagged to the client which sees a 409 error
> >>> message, but non of the really useful output that the hook generates.
> >>> The error text is logged in the apache error_log though:
> >>>
> >>> [Mon Sep 24 09:25:17 2007] [error] [client xxx.xx.xx.xxx] 'pre-commit'
> >>> hook failed with error output:\nDetected reference to blah blah in
> >>> file.x at line 2\nPlease replace with something else\n [409,
> >>> #165001]
> >>>
> >>> The user at the client has no idea what to do next until we send them
> >>> the error_log message from apache.
> >>>
> >>> I can't determine what the trigger for this behavior is, but it
> >>> doesn't always happen. Is it Subversion or Apache problem? I can
> >>> upgrade either or both if needs be.
> >>>
> >>> Thanks,
> >>>
> >>>
> >> Yes, I've observed this problem as well:
> >>
> >> http://svn.haxx.se/dev/archive-2007-09/0389.shtml
> >>
> >> All I could pin down is that it is only triggered by 1.4 clients. I
> >> can reproduce it (via the test harness I use for my hook scripts),
> >> but I haven't figured out why certain tests trigger it. Given that
> >> I can reproduce, I would be happy to provide more data if this
> >> will help get it fixed.
> >>
> >
> > Your previous report refers to issue 443 which has been resolved as
> > FIXED (to be released in 1.5). This code is currently in trunk. Could
> > you test with a trunk client to see whether this is still an issue?
> >
> Tested with trunk client built from r26763 - this works fine on a
> test where the 1.4.3 client fails every time.
Great! Then it looks like you're being served by 1.5. Unfortunately,
there's so much changed in 1.5 in the DAV library that I can't propose
for backport.
bye,
Erik.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: pre-commit stderr output sometimes lost
Posted by Dave Rodgman <da...@fastmail.co.uk>.
Erik Huelsmann wrote:
> On 9/24/07, Dave Rodgman <da...@fastmail.co.uk> wrote:
>
>> Nick Thompson wrote:
>>
>>> Subversion 1.4.3 on RHEL3 with Apache 2.0.55
>>>
>>> Hi,
>>>
>>> Just occasionally, we see that the error output from pre-commit gets
>>> lost. The error is flagged to the client which sees a 409 error
>>> message, but non of the really useful output that the hook generates.
>>> The error text is logged in the apache error_log though:
>>>
>>> [Mon Sep 24 09:25:17 2007] [error] [client xxx.xx.xx.xxx] 'pre-commit'
>>> hook failed with error output:\nDetected reference to blah blah in
>>> file.x at line 2\nPlease replace with something else\n [409,
>>> #165001]
>>>
>>> The user at the client has no idea what to do next until we send them
>>> the error_log message from apache.
>>>
>>> I can't determine what the trigger for this behavior is, but it
>>> doesn't always happen. Is it Subversion or Apache problem? I can
>>> upgrade either or both if needs be.
>>>
>>> Thanks,
>>>
>>>
>> Yes, I've observed this problem as well:
>>
>> http://svn.haxx.se/dev/archive-2007-09/0389.shtml
>>
>> All I could pin down is that it is only triggered by 1.4 clients. I
>> can reproduce it (via the test harness I use for my hook scripts),
>> but I haven't figured out why certain tests trigger it. Given that
>> I can reproduce, I would be happy to provide more data if this
>> will help get it fixed.
>>
>
> Your previous report refers to issue 443 which has been resolved as
> FIXED (to be released in 1.5). This code is currently in trunk. Could
> you test with a trunk client to see whether this is still an issue?
>
Tested with trunk client built from r26763 - this works fine on a
test where the 1.4.3 client fails every time.
cheers
Dave
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: pre-commit stderr output sometimes lost
Posted by Erik Huelsmann <eh...@gmail.com>.
On 9/24/07, Dave Rodgman <da...@fastmail.co.uk> wrote:
> Nick Thompson wrote:
> > Subversion 1.4.3 on RHEL3 with Apache 2.0.55
> >
> > Hi,
> >
> > Just occasionally, we see that the error output from pre-commit gets
> > lost. The error is flagged to the client which sees a 409 error
> > message, but non of the really useful output that the hook generates.
> > The error text is logged in the apache error_log though:
> >
> > [Mon Sep 24 09:25:17 2007] [error] [client xxx.xx.xx.xxx] 'pre-commit'
> > hook failed with error output:\nDetected reference to blah blah in
> > file.x at line 2\nPlease replace with something else\n [409,
> > #165001]
> >
> > The user at the client has no idea what to do next until we send them
> > the error_log message from apache.
> >
> > I can't determine what the trigger for this behavior is, but it
> > doesn't always happen. Is it Subversion or Apache problem? I can
> > upgrade either or both if needs be.
> >
> > Thanks,
> >
> Yes, I've observed this problem as well:
>
> http://svn.haxx.se/dev/archive-2007-09/0389.shtml
>
> All I could pin down is that it is only triggered by 1.4 clients. I
> can reproduce it (via the test harness I use for my hook scripts),
> but I haven't figured out why certain tests trigger it. Given that
> I can reproduce, I would be happy to provide more data if this
> will help get it fixed.
Your previous report refers to issue 443 which has been resolved as
FIXED (to be released in 1.5). This code is currently in trunk. Could
you test with a trunk client to see whether this is still an issue?
Thanks in advance!
bye,
Erik.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: pre-commit stderr output sometimes lost
Posted by Dave Rodgman <da...@fastmail.co.uk>.
Nick Thompson wrote:
> Subversion 1.4.3 on RHEL3 with Apache 2.0.55
>
> Hi,
>
> Just occasionally, we see that the error output from pre-commit gets
> lost. The error is flagged to the client which sees a 409 error
> message, but non of the really useful output that the hook generates.
> The error text is logged in the apache error_log though:
>
> [Mon Sep 24 09:25:17 2007] [error] [client xxx.xx.xx.xxx] 'pre-commit'
> hook failed with error output:\nDetected reference to blah blah in
> file.x at line 2\nPlease replace with something else\n [409,
> #165001]
>
> The user at the client has no idea what to do next until we send them
> the error_log message from apache.
>
> I can't determine what the trigger for this behavior is, but it
> doesn't always happen. Is it Subversion or Apache problem? I can
> upgrade either or both if needs be.
>
> Thanks,
>
Yes, I've observed this problem as well:
http://svn.haxx.se/dev/archive-2007-09/0389.shtml
All I could pin down is that it is only triggered by 1.4 clients. I
can reproduce it (via the test harness I use for my hook scripts),
but I haven't figured out why certain tests trigger it. Given that
I can reproduce, I would be happy to provide more data if this
will help get it fixed.
Dave
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org