Re: Who from OPNFV can -2/+2 lf-releng possibly blocking all OPNFV jjb verification


 

Hi Cedric and Tomi,

 

Given you seem to be open to this idea and maybe even part of the TSC, would you agree to co-submit with me a request to TSC to open up git repo management  permissions to PTLs (branch management (create/delete/…).

This seems to be the case for other LF projects so should not be too difficult to get accepted.

 

Thanks

 

  Alec

 

 

From: <opnfv-tech-discuss@...> on behalf of "Alec via Lists.Opnfv.Org" <ahothan=cisco.com@...>
Reply-To: "Alec Hothan (ahothan)" <ahothan@...>
Date: Wednesday, December 18, 2019 at 9:06 AM
To: Trevor Bramwell <tbramwell@...>
Cc: "opnfv-tech-discuss@..." <opnfv-tech-discuss@...>
Subject: Re: [opnfv-tech-discuss] Who from OPNFV can -2/+2 lf-releng possibly blocking all OPNFV jjb verification

 

Hi Trevor,

 

Inline…

 

 

From: Trevor Bramwell <tbramwell@...>
Date: Tuesday, December 17, 2019 at 3:35 PM
To: <ahothan@...>
Cc: "ollivier.cedric@..." <ollivier.cedric@...>, "opnfv-tech-discuss@..." <opnfv-tech-discuss@...>, "opnfv-tsc@..." <opnfv-tsc@...>
Subject: Re: [opnfv-tech-discuss] Who from OPNFV can -2/+2 lf-releng possibly blocking all OPNFV jjb verification

 

Hey Alec,

 

As far as dockerhub permissions go, I see you should have admin rights

on the nfvbench repository. See attached screenshots.

Though looking at the ticket, Aric may have forgotten to notify you

after the work was done?

 

[alec] Aric did unicast me and I have already (finally) cleaned the repo!

 

 

I don't know the historical context as to why branch creation wasn't

allowed for project owners, but IIRC that was something decided by the

TSC in the past. Lifting that restriction would actually align us in

terms of permissions with other LFN projects.

 

Would you mind adding that request to a future TSC meeting? I can't

imagine there would be much objection.

 

[alec] I sure can but not sure if I need to attend in person at the TSC meeting?

 

 

As far as JJB goes, yeah it's possible to automate branch creation (we

do this for stable branches) and I believe OpenStack even does it this

way for all their branches, but that doesn't mean we have to.

 

[alec] there is value in automating branch creation when they are tied to overall OPNFV release. However there can be multiple good reasons to create branches unrelated to the OPNFV release process and for these cases we should just trust PTLs and let then manage their repo directly.

 

Thanks

 

  Alec

 

 

 

Regards,

Trevor Bramwell

 

On Sun, Dec 15, 2019 at 04:56:20PM +0000, Alec via Lists.Opnfv.Org wrote:

HI Ollivier, thanks for the offer to help. We could try to fix these with jjb but I think the problem is bigger and would need attention from the TSC.

The only reason this dockerhub state is not too much of a problem for the project is that all users currently build their own containers directly (instead of consuming from the dockerhub). However this clearly will not help adoption by new or non expert users.

The docker hub for nfvbench has many issues including:

   *   Tag based container release is no longer working – only latest works

   *   Stable image needs to be removed – it is over 2 years old and no longer updates

   *   Obsolete releases need to be removed

I will also need to create a branch and release images from that branch.

In a nutshell, these are housekeeping tasks that would take me minutes to do if I had the proper permission in the git repo and dockerhub, but would just take too much time and energy to do using jjb (if even possible).

Thanks

   Alec

Date: Sunday, December 15, 2019 at 5:51 AM

Subject: Re: [opnfv-tech-discuss] Who from OPNFV can -2/+2 lf-releng possibly blocking all OPNFV jjb verification

+1.

For the previous release, it was the exact opposite: lots of branches were created without any PTL ack (neither author or reviewer)!

Please send me your ticket. I may be able to help you regarding Dockerhub.

Cédric

On Fri, 2019-12-13 at 22:40 +0000, Alec Hothan (ahothan) wrote:

I would propose that PTLs are given permission to manage the branches for their repos directly. Those who prefer to go through jjb can do so, but we should not force everybody to go thorough JJB.

I have same concern for

·         OPNFV dockerhub repo. I cannot manage my container versions and the ticket I had open for it for months is now closed without any resolution due to the move to LF helpdesk (June 2019, I replied with the information requested but no follow up unfortunately)

·         OPNFV documentation.

We basically need to allow PTL to bypass JJB based workflows which is IMHO very inconvenient and discouraging. How are other LF projects doing?

Thanks

    Alec

From: <opnfv-tech-discuss@...> on behalf of Cedric OLLIVIER <ollivier.cedric@...>

Date: Friday, December 13, 2019 at 2:28 PM

Subject: [opnfv-tech-discuss] Who from OPNFV can -2/+2 lf-releng possibly blocking all OPNFV jjb verification

Hi,

I think a few people have faced with Releng issues for the last days

when creating their stable branches. The problem is linked to releng-

jjb-verify which seems defined out of OPNFV via submodule [1] (hoping

I'm wrong as all OPNFV projects would depend on it) or to the build

host.

    jenkins_jobs.errors.JenkinsJobsException: Unable to lock cache for '

Please don't call reverify/recheck because the second verification

seems incorrect due to wrong triggers and could postpone bigger issues.

Can someone manage this issue?

And who, from OPNFV active members and out of LF, can merge in

lfit/releng-global-jjb?

[1]

Cédric

 

-=-=-=-=-=-=-=-=-=-=-=-

Links: You receive all messages sent to this group.

-=-=-=-=-=-=-=-=-=-=-=-

 

 

Join opnfv-tech-discuss@lists.opnfv.org to automatically receive all group messages.