Make SVN faster for Kubla


Status Graveyard
Server Operations
10 years ago
3 years ago


(Reporter: clouserw, Assigned: oremj)





10 years ago
Kubla maintains three tags of (trunk, stage, and production).  stage and production are updated on cron jobs every 7-15 minutes (not sure of the exact numbers).

Lastely kublad has been tripping over itself and we are getting a lot of "working copy locked" errors.  This happens when an SVN process is already messing with the working copy when another process tries to do something.  

Updating takes so long (often several minutes per tag) that it seems like the copy is locked more often than not.  When this happens people can't use kubla to make changes.

I'm not familiar enough with the setup to make many recommendations but in my experience an `svn up` is essentially disk i/o - are the tags checked out on a local disk (not nfs)?  Also, I hear svn 1.5 is faster but I have no proof.  It looks like we've been upgraing to 1.5 lately on our other boxes - we should make sure whatever is running kubla is up to date too.
I use svn 1.4.6 on my local machine, and it's ridiculously slow, so I have no doubt that kubla is taking forever to update.

Comment 2

10 years ago
[root@dm-cms02 ~]# svn --version
svn, version 1.5.1 (r32289)


Comment 3

10 years ago
Alos, there are no NFS mounts on dm-cms02.


10 years ago
Assignee: server-ops → oremj

Comment 4

10 years ago
The VM only had 512M ram and barely anything was in the FS cache. I doubled the ram.  Please reopen if it still feels too sluggish and we can investigate other options.
Last Resolved: 10 years ago
Resolution: --- → FIXED
Product: → Graveyard
You need to log in before you can comment on or make changes to this bug.