Customize query in gerrit
Details
Details
| Status | Subtype | Assigned | Task | ||
|---|---|---|---|---|---|
| Resolved | • mobrovac | T109926 [keyresult] Migrate all Service team owned services to scap | |||
| Resolved | • mobrovac | T120103 Deploy Parsoid with scap3 | |||
| Resolved | • mobrovac | T90668 Replace custom server.js with service-runner | |||
| Resolved | • GWicke | T135679 RFC: Clarify / clean up return values from ServiceRunner.run() | |||
| Resolved | yuvipanda | T140227 Remove parsoid UID/GID from Labs LDAP | |||
| Resolved | • mobrovac | T140218 Remove automatic beta updates for Parsoid | |||
| Resolved | • mobrovac | T140477 Make Parsoid Vagrant role use service::node | |||
| Resolved | dduvall | T142990 Sequential execution should be per-deployment, not per-phase |
Event Timeline
Comment Actions
deployment-parsoid05 and deployment-cache-parsoid05 seem like likely deployment targets in beta. Looking at the repo_config salt pillar it seems like the transition from trebuchet to scap3 would be pretty straight-forward for beta: checkout submodules, service restart, service checks.
Although, I'm not completely sure these targets use trebuchet. Normally a target has the deploy_target grain set to the repo, would expect: deployment_target: parsoid/deploy to be set on either of those boxes, but I don't see it:
thcipriani@deployment-parsoid05:~$ sudo salt-call grains.get 'deployment_target' local: thcipriani@deployment-salt:~$ sudo salt -G 'deployment_target:parsoid/deploy' test.ping thcipriani@deployment-salt:~$
Is trebuchet used in beta for parsoid?
Comment Actions
Parsoid is the one of the two backend services magically updated via Jenkins (the other is cxserver).
The job is https://integration.wikimedia.org/ci/job/beta-parsoid-update-eqiad/ triggered on post merge of mediawiki/services/parsoid/deploy as well as from the source repo mediawiki/services/parsoid (for a setting file iirc).
The JJB definition is under https://github.com/wikimedia/integration-config/blob/master/jjb/beta.yaml The job runs on the instance hosting the service it:
- clones the repos in a workspace
- rsync them to /srv/deployment/parsoid
- restart the service
The reason for deploy_target to be set is probably because we reuse the same puppet class as on production. So that would set the salt grains for integration-saltmaster.
We might had some repositories using Trebuchet. I remember Ryan made an attempt to switch mediawiki/core to use a couple years or so ago.
Comment Actions
Note that this is a class of tasks mentioned by RelEng in the SoS today; it's (one of several tasks) blocking the scap3 migration.
Comment Actions
Yes, we are first going to migrate to service-runner ( https://gerrit.wikimedia.org/r/#/c/289508/ ) and then rely on service-runner deployment to ease the migration to scap3.
Comment Actions
Change 304468 had a related patch set uploaded (by Mobrovac):
Add the Scap3 configuration
Comment Actions
Change 304470 had a related patch set uploaded (by Mobrovac):
Parsoid: Switch to Scap3 deployments
Comment Actions
Change 304471 had a related patch set uploaded (by Mobrovac):
Admin: Add Parsoid deployers to the deploy-service group
Comment Actions
Change 304471 merged by Alexandros Kosiaris:
Admin: Add Parsoid deployers to the deploy-service group
Comment Actions
Change 307351 had a related patch set uploaded (by Mobrovac):
Depool and repool nodes and set the group size to 7
Comment Actions
Change 307351 merged by jenkins-bot:
Depool and repool nodes and set the group size to 7
Comment Actions
The move has been completed. As of now, deploying Parsoid involves using Scap3. Resolving.