Just a reminder ticket.
This requires: 1. update libtaskotron and resultsdb rpms 2. add namespace key with value of qa to all our tasks: rpmlint, depcheck and upgradepath and update git checkouts on buildmaster 3. run alembic script to update resultsdb schema 4. add namespaces.conf config file to clients via ansible 5. change task names in resultsdb (rpmlint to qa.rpmlint, ...)
libtaskotron
resultsdb
namespace
qa
rpmlint
depcheck
upgradepath
namespaces.conf
qa.rpmlint
I think that should do. If not, put additional info in comments, thanks.
This ticket had assigned some Differential requests: D820
So after a few bumps, it's deployed in dev, for reference linking what needed to be done/fixed:
https://bitbucket.org/fedoraqa/task-rpmlint/commits/90a529a5ae24a1bde19c3babed5ba05fb9139e98 https://bitbucket.org/fedoraqa/task-depcheck/commits/5f6455c63ed36838716c664f558a8bc245bb1551 https://bitbucket.org/fedoraqa/task-upgradepath/commits/255dffcc3e6faa1dcb818734bac0a7377975702b https://bitbucket.org/fedoraqa/task-dockerautotest/commits/4bf9e5f7558ec2a4e1a5c0880df8973089bf3527
https://infrastructure.fedoraproject.org/cgit/ansible.git/commit/?id=0cd311abd967e5c1914e66b38b95b4d9d672d9f7 https://infrastructure.fedoraproject.org/cgit/ansible.git/commit/?id=716f09ed29e19871cfb416400f21c540bb5e5966 https://infrastructure.fedoraproject.org/cgit/ansible.git/commit/?id=8da598eb43d31a4a12fc7de14311d7b2450414c0 https://infrastructure.fedoraproject.org/cgit/ansible.git/commit/?id=a55b3899de2c2cb91d3e6e706c58dba9ca0e057b
Bug found, patch posted: D820
libtaskotron updated to 0.4.11 and resultsdb to 1.1.16 and alembic upgrade run.
Updated testcase names in resultsdb_dev db:
update testcase set name='qa.upgradepath' where name='upgradepath'; update testcase set name='qa.depcheck' where name='depcheck'; update testcase set name='qa.rpmlint' where name='rpmlint';
There are couple of results with old names that were caused by image not being updated and jobs were failing, but it's dev db, I am not going to fix them.
That should be all.
It sounds like doing a test run with a dump of prod might be a good idea before doing the actual deployment, then
! In #756#10669, @tflink wrote: It sounds like doing a test run with a dump of prod might be a good idea before doing the actual deployment, then
Yeah we can, but I think I covered most of the issues (D820 and not using old image) and if I did, updating stg will confirm that :)
This has been deployed to dev, stg and prod. Closing, if there are any issues please open a new ticket.
Metadata Update from @mkrizek: - Issue tagged with: infrastructure
Log in to comment on this ticket.