[SAC] SVN backups and poor interactions was: MapGuide SVN Down Again

[Howard Butler]

Fixed.

After some investigation, it appears that the backup scripts are  
causing the issue.  We had backup scripts being run out of /home/ 
sbarnes/scripts that were hung, then root came along and ran its own  
(and hung).  It appears interaction of these scripts is locking the  
repository and causing our issues.

Shawn,

Can you investigate why crons might still be running out of your home  
directory?  Here's what the process table looked like when I checked  
earlier tonight:

> [hobu at osgeo1 repos]$ ps aux|grep svn
> root      8553  0.0  0.0  2280  880 ?        Ss   15:45   0:00 /bin/ 
> bash /root/scripts/backup_trac_svn.sh
> root      8776  0.0  0.0  3084  868 ?        S    16:01   0:00 sh - 
> c svnadmin dump --quiet --deltas /var/www/svn/repos/mapguide > "/ 
> home/back/svn_backup/2007-02-14_15.45.02"/mapguide
> root      8777  0.0  0.1  8628 2688 ?        S    16:01   0:00  
> svnadmin dump --quiet --deltas /var/www/svn/repos/mapguide
> root     11503  0.0  0.0  2792  868 ?        Ss   18:45   0:00 /bin/ 
> bash /root/scripts/backup_trac_svn.sh
> root     11504  0.0  0.1  4460 2312 ?        S    18:45   0:00  
> perl /home/sbarnes/scripts/svn_repos_backup.pl
> root     11907  0.0  0.0  2756  868 ?        S    19:01   0:00 sh - 
> c svnadmin dump --quiet --deltas /var/www/svn/repos/mapguide > "/ 
> home/back/svn_backup/2007-02-14_18.45.01"/mapguide
> root     11908  0.0  0.1  9416 2792 ?        S    19:01   0:00  
> svnadmin dump --quiet --deltas /var/www/svn/repos/mapguide
> hobu     13533  0.0  0.0  5616  652 pts/0    S+   20:58   0:00 grep  
> svn

I think some of our repositories are quite large to be doing full  
dumps as frequently as we are (~8 times per day).  Gzipped, the  
mapguide repository is ~710mb, and doing a full delta dump of  
probably takes quite a while and is stepping on incoming commits.

I propose change things to do incrementals throughout the day at the  
same frequency we are currently doing them and a single full backup  
outside of peak development hours.  Someone please propose a good UTC  
time to do a full dump so as not to step on folks.  We are worldly  
enough to make this somewhat hard ;)



Howard


>
On Feb 14, 2007, at 6:09 PM, Jason Birch wrote:

> Did MapBender get migrated to the alternate data store?  Did it  
> work OK?
>
> -----Original Message-----
> From: sac-bounces at lists.osgeo.org [mailto:sac-bounces at lists.osgeo.org]
> On Behalf Of Robert Bray
> Sent: Wednesday, February 14, 2007 15:57
> To: System Administration Committee Discussion/OSGeo
> Subject: [SAC] MapGuide SVN Down Again
> Importance: High
>
> 2nd time today...help!
>
> Thanks,
> Bob
> _______________________________________________
> Sac mailing list
> Sac at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/sac
> _______________________________________________
> Sac mailing list
> Sac at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/sac
>