subversion-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Alfred von Campe <alf...@von-campe.com>
Subject Re: Update Error over VPN
Date Tue, 10 Mar 2015 11:16:11 GMT
I was going to chime in earlier, but something in Brane’s response finally put me over the
edge.  We are having a very similar error when updating over VPN (the problem does not occur
when in the office):

Summary of conflicts:
  Text conflicts: 1
svn: E175002: REPORT request on '/hepd/Shelby/!svn/me’ failed

The issue occurs while trying to do a merge, and it happens at the point in time when a few
large binaries (which take noticeably longer when doing an “svn up”  because of their
size) are processed.  The reason I am replying is that we do have a load balancer in front
of our pair of Subversion servers.  I hope this helps to shed some light on this issue.

Alfred



> On Mar 10, 2015, at 2:09, Branko Čibej <brane@wandisco.com> wrote:
> 
> On 10.03.2015 01:09, Stümpfig, Thomas wrote:
>>  
>> Hi Bert, All
>> “a REPORT require on ‘/svn/…./!svn/me’ failed “ this is the only message
my colleague is getting. Same message with command line tools , 
> 
> I'm sorry, but there's no string like that in the Subversion 1.8.x source code. So this
can't be the exact error message.
> 
>> . Server side (VisualSVN) error cannot exactly be linked to the users action –perhaps
this could be an enhancement to mod_dav_svn – send a transaction id in the http header and
log it. The server Error very probably is: "Error writing base64 data: APR does not understand
this error code  [500, #620018] (We have ~300 commits per day and many more reading operations.
>> The error occurs only in WAN circumstances, and only on large updates (>2GB).
In LAN environment everything is ok. The user in question has got a 100Mb/s DOCIS3.0 line
IPV6 based IPV4 carrier grade nat accessing the companies network via juniper client.
> 
> The error code (620018) is the canonical APR error code for "connection aborted", but
the question is, what is terminating the connection (it's also strange that "APR does not
understand that error code," but that's likely a different issue).
> 
> So, do you have a (hardware or software) proxy/load balancer or similar on the entry
point from the WAN? If yes, does it have logs you could look at?
> 
>>  Meanwhile I read about apache timeout,… and set this to 1800 -> 30min… I
hope this fixes the issues …. I’ll keep this list informed
> 
> FWIW, the kind of update you're talking about shouldn't take more than 10 minutes in
the worst case over a 100Mb/s link.
> 
> -- Brane
> 


Mime
View raw message