Closed Bug 1314688 Opened 8 years ago Closed 8 years ago

Decommission reviewboard-hg1

Categories

(MozReview Graveyard :: Infrastructure, defect)

Product:
MozReview Graveyard
Component:
Infrastructure
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

(Not tracked)

Status:
RESOLVED WONTFIX

People

(Reporter: gps, Unassigned)

References

Details

We just switched from reviewboard-hg1 to reviewboard-hg2. reviewboard-hg1 can be decomissioned (after a cool down period to ensure we don't need to revert to it, of course).

I'll let fubar give the green light for action here.
fubar: can we decom this host? If not fully, perhaps we can power it down so it stops providing noise in e.g. security scans.
Flags: needinfo?(klibby)
Should we re-kick it as CentOS 7 with an eye toward making reviewboard-hg more like hgmo? Your note to Jake about quantum and autoland make me a little twitchy about this being a SPOF.
Flags: needinfo?(klibby) → needinfo?(gps)
If we're talking about making reviewboard-hg more like hgmo, honestly I'd rather do away with reviewboard-hg and use hgmo! But this is... complicated.

I am all for removing the SPOF from reviewboard-hg. Doing that will require moving the hg repos to off-host storage (like NFS). And this includes the HTTP endpoint. And last time we had multiple HTTP servers hitting NFS, we ran into weird perf issues. While I think most of those have been fixed in Mercurial upstream, it's a risk. I suppose we could forego having multiple hosts in the ZLB and use a hot standby like we have hgssh configured. We would still need to move to NFS or something.

So, uh, we should do something. But I'm not sure what the timeline will be. Since VMs can be created easily, perhaps we should get reviewboard-hg1 reimaged as CentOS 7 so it is available in case disaster strikes? We can worry about larger architecture changes later?
Flags: needinfo?(gps)
(In reply to Gregory Szorc [:gps] from comment #3)
>
> Since VMs can be created easily, perhaps we should get reviewboard-hg1
> reimaged as CentOS 7 so it is available in case disaster strikes? We can
> worry about larger architecture changes later?

That's pretty much exactly what I was thinking. I'll file a bug with IT and wontfix this one.
Status: NEW → RESOLVED
Closed: 8 years ago
Resolution: --- → WONTFIX
You need to log in before you can comment on or make changes to this bug.