[Buildbot-devel] Scheduler maintenance windows?

[Scott Lamb]

Update on this: the Perforce people gave us a complex procedure[*]  
for checkpointing without locking the server for an hour, so once we  
get it implemented, we won't need any buildbot workarounds. Not sure  
if this is a feature anyone else wants or not.

[*] - Essentially, you bootstrap a second server by copying a  
checkpoint over from the primary. Then periodically you get new  
checkpoints by rolling the journals on your primary server, copying  
the freshly-rolled journals to the second server, replaying them,  
then checkpointing there. Latest second server checkpoint + current  
primary server journal = current primary server state. In effect,  
it's the same as just checkpointing the primary one, but you never  
took it down and you might end up a lot more confused. It works if  
you're careful to get the details right.

On Apr 25, 2006, at 12:53 AM, Scott Lamb wrote:

> I'm considering a patch to Scheduler to make it avoid scheduling  
> anything during a specific window every night. Our Perforce server  
> gets backed up nightly and runs a checkpoint, which *locks* the  
> database for a ridiculous long time - something like an hour. (The  
> Perforce people apparently don't believe in MVCC.) Builds during  
> this time fail like this:
>
>     starting p4
>     ...
>     command timed out: 1200 seconds without output, killing pid 5883
>     program finished with exit code -1
>
> I suppose I could crank up the timeout on that step a *lot*, but I  
> think it's better to just queue builds until the known window has  
> ended. Either by time or by the Perforce server's pre- and post- 
> backup scripts notifying buildbot in some way.
>
> Thoughts?
>
> -- 
> Scott Lamb <http://www.slamb.org/>
>
>
>
>
> -------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services,  
> security?
> Get stuff done quickly with pre-integrated technology to make your  
> job easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache  
> Geronimo
> http://sel.as-us.falkag.net/sel? 
> cmd=lnk&kid=120709&bid=263057&dat=121642
> _______________________________________________
> Buildbot-devel mailing list
> Buildbot-devel at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/buildbot-devel

-- 
Scott Lamb <http://www.slamb.org/>