Opened 6 weeks ago

Closed 22 hours ago

#33922 closed task (fixed)

hardware requirements planning for gitlab launch

Reported by: anarcat Owned by: hiro
Priority: Medium Milestone:
Component: Internal Services/Services Admin Team Version:
Severity: Normal Keywords: tpa-roadmap-may
Cc: Actual Points:
Parent ID: #29400 Points:
Reviewer: Sponsor:

Description

we need to make sure our gitlab instance will survive the migration from trac. so far it's working well, but it might not be happy with the extra traffic that Trac redirects will bring.

talk to other gitlab instances to see how much traffic they get and how they handle it.

Child Tickets

Change History (6)

comment:1 Changed 6 weeks ago by anarcat

Keywords: tpa-roadmap-april added
Owner: set to hiro
Status: newassigned

comment:2 Changed 3 weeks ago by hiro

Resolution: fixed
Status: assignedclosed

comment:3 Changed 3 weeks ago by hiro

Resolution: fixed
Status: closedreopened

comment:4 Changed 3 weeks ago by hiro

Keywords: tpa-roadmap-may added; tpa-roadmap-april removed

comment:5 Changed 3 weeks ago by anarcat

just to be clear here. we're happy with the gitlab hardware right now, pre-launch, that is true.

but we should have estimates on the hardware requirements of the project *after* launch. specifically, we need to take into account the fact that we might have way more git repositories on the server during and after launch. we should plan to have similar I/O capacity than what cupani *AND* troodi currently have, if not more.

this should be done by looking at Grafana graphs of disk, memory and CPU usage and see if we have the capacity to provide for those.

we should also make it clear that we are not replacing Jenkins in this phase, for example. or that we do *not* plan on providing gitlab pages services. in fact, it might be worth going through the GitLab feature set and make sure we have a yes/no answer for each service, and to make sure we have the hardware capacity to respond to the "yes" services.

i'm particularly concerned about git repositories, in other words, but there might be other thing that I'm missing, and this ticket should answer that in a definite way, by setting expectations for the team and hardware requirements for TPA.

thanks!

comment:6 Changed 22 hours ago by anarcat

Resolution: fixed
Status: reopenedclosed

so far, we're going to throw hardware at this problem if we have any performance related issues. we have room for this in our budget and ops would probably give us budget for this migration if we need it.

so let's assume we can just throw one more ganeti node at the problem if it occurs.

Note: See TracTickets for help on using tickets.