Modify the puppet logic which browserid applications follow in the following ways : * assert a desired browserid-server rpm version in either a variable or a heira key/val setting * if the desired version is different than the current version then ** attempt to drain the client from zues using the broker (Bug 783501) *** if the broker won't drain the client because there isn't enough capacity then maybe loop and re-attempt to drain or just skip doing the update until the next puppet run *** elseif the broker claims the client is currently draining then it's likely we're just waiting for the drain to complete before continuing the upgrade so we should just skip doing the updat until the next puppet run *** elseif the broker says the client is drained then we continue ** yum upgrade the rpm to the desired version ** restart daemontools services Also, if there have been nginx changes, nginx should restart https://intranet.mozilla.org/Services/Ops/BrowserID/CodePush#Process_in_code
This depends on the work in Bug 783506 in order to be able to get the rpm via yum
I'll be testing this in staging today. Initially on web1.idweb.scl2.stage.svc.mozilla.com
no longer relevant
Status: NEW → RESOLVED
Last Resolved: 5 years ago
Resolution: --- → INVALID
You need to log in before you can comment on or make changes to this bug.