> In addition, please bear in mind that landing bustage on trunk trees actually > makes the Try wait times worse (since the trunk backouts/retriggers take test > job priority over Try) - leading to others not bothering to use Try either, > and so > the situation cascades.
I thought tryserver used a different pool of machines isolated from all the other trees, because we treated the tryserver machines as pwned. Is that not or no longer the case? > Increasingly over the last few weeks, people have been landing untested > changes on mozilla-inbound that has resulted in bustage - which with high > levels of coalescing (that makes finding the culprit require multiple > time-consuming re-triggers) Is there a plan to mitigate the coalescing on m-i? It seems like that is a big part of the problem. On Thu, Aug 9, 2012 at 10:19 AM, Ed Morley <bmo.takethis...@edmorley.co.uk> wrote: > Increasingly over the last few weeks, people have been landing untested > changes on mozilla-inbound that has resulted in bustage - which with high > levels of coalescing (that makes finding the culprit require multiple > time-consuming re-triggers) has meant closing the tree for several hours. > > For example - last night the tree was closed for a total of roughly 6 hours > whilst 7 backouts were performed. This makes three days in a row where the > tree has had to be closed during peak times. > > This is neither fair on other developers (who are understandably frustrated > when the tree is closed day after day), nor on the sheriffs (whom at that > time of day are only volunteers). > > As such, I would like to request that people adjust their risk thresholds for > when to use Try - so we can minimise the impact on the tree. > > I realise that Try Server end-to-end times have been less than ideal > recently, so using try can be frustrating - however the situation is slowing > improving (tegra reboot issues fixed, pymake for windows almost ready, > see[/ping] bug 772458 for more). In addition, please bear in mind that > landing bustage on trunk trees actually makes the Try wait times worse (since > the trunk backouts/retriggers take test job priority over Try) - leading to > others not bothering to use Try either, and so the situation cascades. > > A quick reminder: If your patch(es) have been sent to try, please add the URL > to the bug, so in the case of bustage, it's easier for the sheriffs to > eliminate your push as the cause. > > Finally, the list of steps to perform after pushing to inbound has recently > been simplified (thanks to the new merge tool we use) - please take a look if > you haven't in the last 4-6 weeks: > https://wiki.mozilla.org/Tree_Rules/Inbound#Please_do_the_following_after_pushing_to_inbound > > Best wishes, > > Ed > _______________________________________________ > dev-platform mailing list > dev-platform@lists.mozilla.org > https://lists.mozilla.org/listinfo/dev-platform _______________________________________________ dev-platform mailing list dev-platform@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-platform
