Closed rehandalal closed 7 years ago
Have you talked to @milescrabill about the cron setup? In Normandy we opted to just define the management commands, and leave cron scheduling up to the hosting environment, which has been helpful.
In the tests, it looks like you need to update the object factories as well?
What is the intention behind the cron command? We hit an endpoint via cron from the host instead. I would advise keeping all state in a DB (and none on the host/container), though an in-memory buffer would work depending on use case.
Tests will probably fail due to issues fixed in #47
@milescrabill Although there is a cron job that hits the webheads once a minute, that is for health checks, and more importantly, happens once per webhead. The jobs here need to happen once per day regardless of the number of webheads. It would be a lot more robust if we just use cron.
In Normandy we have cron externally managed in Puppet, and it calls the needed commands on the needed schedule. I think that is defined here for Normandy.
@relud does that summarize Normandy's cron correctly?
yep
That arrangement is what Rehan and I discussed yesterday - the summary was that he would tell me the endpoint / command that needs to be hit / run and I would write the relevant Puppet code. I was referencing Normandy's Puppet code in that conversation, but I didn't realize you (the devs) could see it.
On Fri, Oct 14, 2016 at 1:19 PM, Daniel Thorn notifications@github.com wrote:
yep
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/mozilla/morgoth/pull/39#issuecomment-253909036, or mute the thread https://github.com/notifications/unsubscribe-auth/AEOcLKxrpj6f96p3R3YPRKemCGGqVGqGks5qz-O2gaJpZM4KJDb9 .
r?