View source for Balrog/Meetings/CloudOps - July 27, 2016

= Attendees =
bhearsum, ckolos, mostlygeek

= Notes =
* New Docker repos for Balrog Agent & Worker
* Try running cleanup cronjob?
** Wait until Ben is back from PTO, second week of August
* How/Where are schema upgrades done?
** Schema changes need to be compatible with current version of code + new version.
** Schema upgrades can be done prior to pushing the new code, or after. This should be specified in the bug.
* How to request that pub/admin are always deployed together?
** Benson is going to make some changes to CloudOps process to make this easier; we plan to do it for all future pushes.
** Pub and admin don't have to be deployed at exactly the same moment, just as part of the same push (hopefully within ~1h of one another).
* bhearsum mentioned that we're looking at enabling native table compression for releases_history
** ckolos said we should make sure that RDS supports it before going any further
** Benson says we should snapshot prod db and see how long it will take, and see if there's any other downsides
** Will be talked about more before the change is landed.
* Discussion of escalation paths from https://wiki.mozilla.org/Balrog#Support_.26_Escalation - CloudOps suggested that we set-up a pagerduty email address to make it easier to escalate back through RelEng when necessary. E-mail to follow.
* Can we get stage hooked up to LDAP?
** Bad idea, because more people have access to stage infra than prod, and we don't want that set of people to have access to LDAP.
** Alternative: Ask for new accounts to get added to stage whenever necessary (should do this at the same time as requesting access to prod for people). If this gets too burdensome Benson may try to hack something up, it will probably be fine though.