Hi Yuri,
On 14/02/2015 17:22, Yuri Weinstein wrote:
>> Yeah. Well, the last run alone isn't so important; we want to see a
>> string of clean runs because a lot of issues aren't reproduced in
>> every run.
>
> My hope was that we can see all "green" results for say this giant release/backport, but I agree that we would need to make our go/no-go decision based on multiple run results, as I am not sure if we can get them all "green" due to complexity, time needed to execute, environment state etc..
>
> We could thou modify our process a bit:
> 1. after backport-branch is ready for QE, merge it to the named branch (say 'giant' in this example) - that what we did now
> 2. cut a release numbered brach (maybe it's tag, not sure), say "v0.87.1"
> 3. run all QE suites on "v0.87.1" and get it to "all passed" state
> 4. make sure that commits to "v0.87.1" are committed to the named branch ('giant')
That makes sense to me, only with s/v0.87.1/78c71b9200da5e7d832ec58765478404d31ae6b5/.
> #2 is that we have not done this time.
We have not done #2 but we have cut the branch at given SHA ( 78c71b9200da5e7d832ec58765478404d31ae6b5 ) instead, which is can be referenced by a tag if and when it is released. In the mail "Re: giant integration branch for v0.87.1 ready for QE" dated 11th february 2015 I wrote:
> The giant-backports pull requests were merged into https://github.com/ceph/ceph/tree/giant which is not ready for testing.
> For the record, the head is https://github.com/ceph/ceph/commit/78c71b9200da5e7d832ec58765478404d31ae6b5
We cannot add a v0.87.1 tag to the branch before the release process is complete because we won't be able to change it afterwards (people rely on the fact that the history of the giant branch is not rewritten and that tags references are not changed). If during the QE test process we discover that a backport must be included (I'm thinking about https://github.com/ceph/ceph/pull/3731 for instance), 78c71b9200da5e7d832ec58765478404d31ae6b5 won't be v0.87.1 after all.
In a nutshell I think we're having the same view of the process, modulo the timing of the tagging of the release.
Cheers
>
> Thx
> YuriW
>
> ----- Original Message -----
> From: "Gregory Farnum" <greg@xxxxxxxxxxx>
> To: "Loic Dachary" <loic@xxxxxxxxxxx>
> Cc: "Ceph Development" <ceph-devel@xxxxxxxxxxxxxxx>
> Sent: Friday, February 13, 2015 11:56:18 PM
> Subject: Re: [Ceph-qa] 1 hung, 11 passed in teuthology-2015-02-11_16:13:01-samba-giant-distro-basic-multi
>
> On Fri, Feb 13, 2015 at 10:34 PM, Loic Dachary <loic@xxxxxxxxxxx> wrote:
>> Hi Greg,
>>
>> I'm curious to know how you handle the flow of mails from QA runs. Here is a wild guess:
>>
>> * from time to time check that the nightlies run the suites that should be run
>
> Uh, I guess?
>
>> * read the ceph-qa reports daily
>
> Yeah
>
>> * for each failed job, either relate it to an issue or create one or declare it noise
>
> Yeah
>
>> * if a job fails on an existing ticket store a link to the job if it's rare occurrence and the cause is not yet known
>
> Yeah, or just to make clear it's still happening or whatever
>
>> * bi-weekly bug scrub makes sure no issue, old or new, is forgotten
>
> Hopefully!
>
>> * at release time you decide that it is ready based on:
>> ** the list of urgent/immediate issues that you can browse to ensure no issue is a blocker
>> ** the last run of each suite to ensure they are recent enough and environmental noise did not permanently shadow anything
>
> Yeah. Well, the last run alone isn't so important; we want to see a
> string of clean runs because a lot of issues aren't reproduced in
> every run.
> -Greg
> --
> To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at http://vger.kernel.org/majordomo-info.html
>
--
Loïc Dachary, Artisan Logiciel Libre
Attachment:
signature.asc
Description: OpenPGP digital signature
