Scap, obviously, used to deal with multidatacenters (eqiad+tampa), but the way it was implemented caused performance issues (when we removed tampa from the scap list the time to scap dropped a ton).
Let's be prepared for the Dallas DC and not annoy deployers with longer deploys.
Version: wmf-deployment
Severity: normal
| Status | Subtype | Assigned | Task | ||
|---|---|---|---|---|---|
| Resolved | None | T71572 [scap] multi datacenter aware without (major) performance hit | |||
| Resolved | Dzahn | T95436 install/deploy mira as codfw deployment server | |||
| Resolved | RobH | T91678 setup deployment server in codfw (tin equivalent) | |||
| Resolved | Papaul | T95437 server mira setup/mgmt dns/label/port info | |||
| Resolved | Dzahn | T99919 need package 'trebuchet-trigger' for trusty | |||
| Resolved | • chasemp | T99928 errors reported by "eventual_consistency_deployment_server_init" on new deploy server | |||
| Resolved | • demon | T104826 [scap] Add support for syncing /srv/mediawiki-staging including fully working git data to warm spare deploy server | |||
| Resolved | Dzahn | T119165 l10nupdate user uid mismatch between tin and mira | |||
| Resolved | MoritzMuehlenhoff | T113351 Ferm rules for tin/mira | |||
| Resolved | Dzahn | T115075 ssh from tin to mira broken |
Closing as we didn't hear much in the way of yells after we added codfw. If there are specific things we can do to make scap faster we can open other tasks for them (there's already some).
We did do a couple of things for this that weren't tracked by this task: