Replace slaves (Linux Precise -> Trusty) on tryserver.blink. |
|||||||
Issue descriptionCurrent configuration: slave501-c4 slave502-c4 slave503-c4 slave504-c4 slave505-c4 are used for: linux_precise_blink_compile_dbg linux_precise_blink_compile_rel linux_precise_blink_dbg linux_precise_blink_rel and slave137-c4 is used for linux_trusty_blink_rel. We want to remove all *precise* builders and have at least 5 slaves allocated for linux_trusty_blink_compile_dbg linux_trusty_blink_compile_rel linux_trusty_blink_dbg linux_trusty_blink_rel This is required as part of dropping Precise support for layout tests ( bug 660580 ). Since I changed CQ_INCLUDE_TRYBOTS to linux_trusty_blink_rel before making this change, which in retrospect was the wrong order, linux_trusty_blink_rel is a bit overworked right now :-/ (bug 665246)
,
Nov 16 2016
Thanks for taking the initiative on the Precise -> Trusty upgrade I could be wrong but we can just respawn your Precise instance to Trusty and there wouldn't be any hostname change. There will be a short downtime if you're ok with that.
,
Nov 16 2016
That sounds good to me :-)
,
Nov 16 2016
Whatever gets the CQ back to happy the fastest SGTM. If respawning the instances takes much time at all, we should modify git-cl upload and the recipes to stop adding that builder temporarily.
,
Nov 16 2016
Alright, so are the slaves being respawned now? If so, then I assume it's OK for me to commit the change to master.tryserver.blink/slaves.cfg and restart the master?
,
Nov 16 2016
Ah, and another question: Would it be better to keep slave137-c4 and allocate it to also be used for linux_trusty_blink_compile_dbg linux_trusty_blink_compile_rel linux_trusty_blink_dbg linux_trusty_blink_rel (Since linux_trusty_blink_rel needs a bit more capacity than the other try bots on tryserver.blink?) Or maybe later if we decide we need more capacity we'll add more later, and for now we'll stop using slave137-c4?
,
Nov 17 2016
Sorry for the delay. When do you want the slaves respawned? I don't know anything about your builders, I just manage the bots.
,
Nov 17 2016
You can just put that one bot on the other builders if you want...
,
Nov 17 2016
Sorry for not responding myself. We needed to re-land a couple of CLs to hopefully prevent mishaps. We can respawn them tomorrow. Downtime should not be an issue.
,
Nov 17 2016
So, in the meantime the v8 auto-roller was switched to use those trybots again and they are broken ATM: https://build.chromium.org/p/tryserver.blink/builders/linux_precise_blink_rel/builds/5201 I'll skip them manually on the rolls until tomorrow.
,
Nov 17 2016
You want me to do this now?
,
Nov 17 2016
Clarification: Do you want me to respawn slave{{501..505},137}-c4 now?
,
Nov 17 2016
Yep, I think now would be good :-) -- we're expecting just a little downtime for those slaves while they restart, right?
,
Nov 17 2016
+ current Chromium sheriffs: This changes means that all of the linux builders on chromium.webkit will be offline for a short time.
,
Nov 18 2016
slave{{501..505},137}-c4 have been respawned.
,
Nov 18 2016
Excellent, thanks :-) Since the slaves are updated, that means this issue is done; more updates on the main issue ( issue 660580 ).
,
Nov 23 2016
Looks like four of the slaves of linux_trusty_blink_rel are running Precise (os version = 12.04): https://build.chromium.org/p/tryserver.blink/builders/linux_trusty_blink_rel https://build.chromium.org/p/tryserver.blink/buildslaves/slave501-c4 https://build.chromium.org/p/tryserver.blink/buildslaves/slave502-c4 https://build.chromium.org/p/tryserver.blink/buildslaves/slave503-c4 https://build.chromium.org/p/tryserver.blink/buildslaves/slave504-c4 https://build.chromium.org/p/tryserver.blink/buildslaves/slave505-c4 We might have to try upgrading/respawning these again?
,
Nov 23 2016
Sorry about that. Just respawned them.
,
Nov 23 2016
|
|||||||
►
Sign in to add a comment |
|||||||
Comment 1 by qyears...@chromium.org
, Nov 16 2016