[Gluster-devel] [Gluster-infra] freebsd smoke failure

[Atin Mukherjee]

On 01/12/2016 04:10 PM, Rajesh Joseph wrote:
> 
> 
> ----- Original Message -----
>> From: "Niels de Vos" <ndevos at redhat.com>
>> To: "Rajesh Joseph" <rjoseph at redhat.com>
>> Cc: "Atin Mukherjee" <amukherj at redhat.com>, "Gluster Devel" <gluster-devel at gluster.org>, "Gluster Infra"
>> <gluster-infra at gluster.org>
>> Sent: Tuesday, January 12, 2016 2:48:36 PM
>> Subject: Re: [Gluster-devel] [Gluster-infra] freebsd smoke failure
>>
>> On Tue, Jan 12, 2016 at 03:52:43AM -0500, Rajesh Joseph wrote:
>>>
>>>
>>> ----- Original Message -----
>>>> From: "Niels de Vos" <ndevos at redhat.com>
>>>> To: "Atin Mukherjee" <amukherj at redhat.com>
>>>> Cc: "Gluster Devel" <gluster-devel at gluster.org>, "Gluster Infra"
>>>> <gluster-infra at gluster.org>
>>>> Sent: Tuesday, January 12, 2016 1:37:50 PM
>>>> Subject: Re: [Gluster-devel] [Gluster-infra] freebsd smoke failure
>>>>
>>>> On Tue, Jan 12, 2016 at 11:19:30AM +0530, Atin Mukherjee wrote:
>>>>> I've been observing freebsd smoke failure for all the patches for last
>>>>> few days with the following error:
>>>>>
>>>>> mkdir: /usr/local/lib/python2.7/site-packages/gluster: Permission
>>>>> denied
>>>>> mkdir: /usr/local/lib/python2.7/site-packages/gluster: Permission
>>>>> denied
>>>>>
>>>>> Can any one from infra team can help here?
>>>>
>>>> This is not an infra issue, it is a build/installation issue in Gluster
>>>> that the infra caught.
>>>>
>>>> http://review.gluster.org/13208 should fix it, but I was at loss why
>>>> only smoke tests got triggered and regression tests not.
>>>>
>>>
>>> I am just wondering why are we seeing this smoke failure now and not on
>>> earlier patches?
>>> Is it a regression? I believe smoke was passing sometime back.
>>
>> I think/suspect that the FreeBSD slaves are now managed by Salt and
>> manual changes might have been undone. If people suggested (like now)
>> that this was an issue on the slave, one of the admins might have opened
>> up the permissions on the directories on request.
>>
>> Things like this are really bugs in the build system though.
>>
>>> Anyway, if this is the fix then can we fast track the patch merge?
>>> Thanks Niels for sending the patch.
>>
>> Well, the failing of the smoke test does not seem to add Verified=-1
>> anymore (no idea why)? Whats the reason of the hurry?
> 
> I was in the impression that all smoke runs should pass. I used to check
> result of smoke test before merging any patch. I guess I am wrong.
No, that should be the criteria and as Niels pointed out if it fails
then verified should be set to -1. We do have cases where the regression
overwrites it, but in one my patch regression was never run but still it
didn't set the flag to -1.
>