Disable 'required status check' for master
Hi, We have been having issues getting CI to work fast enough. Currently the CI is being the bottleneck to get commits in. This is severely impacting development. I propose we disable 'required status check' for the master branch so that we can development is not impacted. We can work on fixing the CI situation in parallel. Committer, Please vote if you are okay with disabling 'required status check' for master. Once we have enough votes, I'll request a mentor to file a ticket with infra. Thanks, Indu
Re: Disable 'required status check' for master
This makes sense to me. CI isn't of value if it isn't continuous. +1. On Tue, Oct 31, 2017 at 12:07 PM, Indhu wrote: > Hi, > > We have been having issues getting CI to work fast enough. Currently the CI > is being the bottleneck to get commits in. This is severely impacting > development. I propose we disable 'required status check' for the master > branch so that we can development is not impacted. We can work on fixing > the CI situation in parallel. > > Committer, > Please vote if you are okay with disabling 'required status check' for > master. > > Once we have enough votes, I'll request a mentor to file a ticket with > infra. > > Thanks, > Indu >
Re: Disable 'required status check' for master
BTW, here is a separate thread where there is discussion happening on improving the CI system: https://lists.apache.org/thread.html/13e5ba08acd2f28a49f791c23f4baf97ba9e5bcdabe43aec598d1e19@%3Cdev.mxnet.apache.org%3E While we will eventually get a faster CI system, it is important to disable the checks for now to not block or slow down development until we have a faster CI system. On Tue, Oct 31, 2017 at 12:07 PM Indhu wrote: > Hi, > > We have been having issues getting CI to work fast enough. Currently the > CI is being the bottleneck to get commits in. This is severely impacting > development. I propose we disable 'required status check' for the master > branch so that we can development is not impacted. We can work on fixing > the CI situation in parallel. > > Committer, > Please vote if you are okay with disabling 'required status check' for > master. > > Once we have enough votes, I'll request a mentor to file a ticket with > infra. > > Thanks, > Indu > >
Re: Disable 'required status check' for master
-1 I personally think it's a necessary evil and a good forcing factor to get CI fixed. Before that requirements, things that failed unit tests were being pushed into master daily. It was a big problem. At least now master is stable. On Tue, Oct 31, 2017 at 12:22 PM, Hen wrote: > This makes sense to me. CI isn't of value if it isn't continuous. +1. > > On Tue, Oct 31, 2017 at 12:07 PM, Indhu wrote: > > > Hi, > > > > We have been having issues getting CI to work fast enough. Currently the > CI > > is being the bottleneck to get commits in. This is severely impacting > > development. I propose we disable 'required status check' for the master > > branch so that we can development is not impacted. We can work on fixing > > the CI situation in parallel. > > > > Committer, > > Please vote if you are okay with disabling 'required status check' for > > master. > > > > Once we have enough votes, I'll request a mentor to file a ticket with > > infra. > > > > Thanks, > > Indu > > >
Re: Disable 'required status check' for master
The number of pending PRs is growing very fast. At the current rate it will reach 200 before we can fix jenkins. Stability is not the only goal. Master will be most stable if we don't push anything, but that's not what we want. Committers should be responsible for their commits. Good judgement is the ultimate guarantee of good code. Thanks, Eric On 2017-10-31 12:38, Chris Olivier wrote: > -1 > > I personally think it's a necessary evil and a good forcing factor to get > CI fixed. > Before that requirements, things that failed unit tests were being pushed > into master daily. It was a big problem. At least now master is stable. > > On Tue, Oct 31, 2017 at 12:22 PM, Hen wrote: > > > This makes sense to me. CI isn't of value if it isn't continuous. +1. > > > > On Tue, Oct 31, 2017 at 12:07 PM, Indhu wrote: > > > > > Hi, > > > > > > We have been having issues getting CI to work fast enough. Currently the > > CI > > > is being the bottleneck to get commits in. This is severely impacting > > > development. I propose we disable 'required status check' for the master > > > branch so that we can development is not impacted. We can work on fixing > > > the CI situation in parallel. > > > > > > Committer, > > > Please vote if you are okay with disabling 'required status check' for > > > master. > > > > > > Once we have enough votes, I'll request a mentor to file a ticket with > > > infra. > > > > > > Thanks, > > > Indu > > > > > >
Re: Disable 'required status check' for master
+1 We enabled protected merge to master at a point when CI was stable (>90% build completion). However, with recent degradation in CI setup, it is no more practical. I see too many PRs in pending and development slowing down. Since new CI design and development is in progress and we have "nightly tests running", disabling protected merge would be a better choice till we get stable CI in place. Thanks, Sandeep On Tue, Oct 31, 2017 at 1:17 PM, Eric Xie wrote: > The number of pending PRs is growing very fast. At the current rate it > will reach 200 before we can fix jenkins. > > Stability is not the only goal. Master will be most stable if we don't > push anything, but that's not what we want. > > Committers should be responsible for their commits. Good judgement is the > ultimate guarantee of good code. > > Thanks, > Eric > > On 2017-10-31 12:38, Chris Olivier wrote: > > -1 > > > > I personally think it's a necessary evil and a good forcing factor to get > > CI fixed. > > Before that requirements, things that failed unit tests were being pushed > > into master daily. It was a big problem. At least now master is stable. > > > > On Tue, Oct 31, 2017 at 12:22 PM, Hen wrote: > > > > > This makes sense to me. CI isn't of value if it isn't continuous. +1. > > > > > > On Tue, Oct 31, 2017 at 12:07 PM, Indhu > wrote: > > > > > > > Hi, > > > > > > > > We have been having issues getting CI to work fast enough. Currently > the > > > CI > > > > is being the bottleneck to get commits in. This is severely impacting > > > > development. I propose we disable 'required status check' for the > master > > > > branch so that we can development is not impacted. We can work on > fixing > > > > the CI situation in parallel. > > > > > > > > Committer, > > > > Please vote if you are okay with disabling 'required status check' > for > > > > master. > > > > > > > > Once we have enough votes, I'll request a mentor to file a ticket > with > > > > infra. > > > > > > > > Thanks, > > > > Indu > > > > > > > > > > -- Sandeep Krishnamurthy
Re: Disable 'required status check' for master
I see 73 PR's for mxnet. 17 of those have successfully passed CI and are waiting for merge (assuming the CR passed human inspection). 21 Show some form of CI in progress. *Only 8 are in the queue in Jenkins* The rest have failed CI for one reason or another. On Tue, Oct 31, 2017 at 1:17 PM, Eric Xie wrote: > The number of pending PRs is growing very fast. At the current rate it > will reach 200 before we can fix jenkins. > > Stability is not the only goal. Master will be most stable if we don't > push anything, but that's not what we want. > > Committers should be responsible for their commits. Good judgement is the > ultimate guarantee of good code. > > Thanks, > Eric > > On 2017-10-31 12:38, Chris Olivier wrote: > > -1 > > > > I personally think it's a necessary evil and a good forcing factor to get > > CI fixed. > > Before that requirements, things that failed unit tests were being pushed > > into master daily. It was a big problem. At least now master is stable. > > > > On Tue, Oct 31, 2017 at 12:22 PM, Hen wrote: > > > > > This makes sense to me. CI isn't of value if it isn't continuous. +1. > > > > > > On Tue, Oct 31, 2017 at 12:07 PM, Indhu > wrote: > > > > > > > Hi, > > > > > > > > We have been having issues getting CI to work fast enough. Currently > the > > > CI > > > > is being the bottleneck to get commits in. This is severely impacting > > > > development. I propose we disable 'required status check' for the > master > > > > branch so that we can development is not impacted. We can work on > fixing > > > > the CI situation in parallel. > > > > > > > > Committer, > > > > Please vote if you are okay with disabling 'required status check' > for > > > > master. > > > > > > > > Once we have enough votes, I'll request a mentor to file a ticket > with > > > > infra. > > > > > > > > Thanks, > > > > Indu > > > > > > > > > >
Re: Disable 'required status check' for master
Its the failed ones that's the problem. They fail because CI is not working properly. For example this trivial change to doc cannot be merged because CI failed: https://github.com/apache/incubator-mxnet/pull/8555/files On 2017-10-31 13:36, Chris Olivier wrote: > I see 73 PR's for mxnet. > > 17 of those have successfully passed CI and are waiting for merge (assuming > the CR passed human inspection). > 21 Show some form of CI in progress. > *Only 8 are in the queue in Jenkins* > > The rest have failed CI for one reason or another. > > > > > On Tue, Oct 31, 2017 at 1:17 PM, Eric Xie wrote: > > > The number of pending PRs is growing very fast. At the current rate it > > will reach 200 before we can fix jenkins. > > > > Stability is not the only goal. Master will be most stable if we don't > > push anything, but that's not what we want. > > > > Committers should be responsible for their commits. Good judgement is the > > ultimate guarantee of good code. > > > > Thanks, > > Eric > > > > On 2017-10-31 12:38, Chris Olivier wrote: > > > -1 > > > > > > I personally think it's a necessary evil and a good forcing factor to get > > > CI fixed. > > > Before that requirements, things that failed unit tests were being pushed > > > into master daily. It was a big problem. At least now master is stable. > > > > > > On Tue, Oct 31, 2017 at 12:22 PM, Hen wrote: > > > > > > > This makes sense to me. CI isn't of value if it isn't continuous. +1. > > > > > > > > On Tue, Oct 31, 2017 at 12:07 PM, Indhu > > wrote: > > > > > > > > > Hi, > > > > > > > > > > We have been having issues getting CI to work fast enough. Currently > > the > > > > CI > > > > > is being the bottleneck to get commits in. This is severely impacting > > > > > development. I propose we disable 'required status check' for the > > master > > > > > branch so that we can development is not impacted. We can work on > > fixing > > > > > the CI situation in parallel. > > > > > > > > > > Committer, > > > > > Please vote if you are okay with disabling 'required status check' > > for > > > > > master. > > > > > > > > > > Once we have enough votes, I'll request a mentor to file a ticket > > with > > > > > infra. > > > > > > > > > > Thanks, > > > > > Indu > > > > > > > > > > > > > > >