446401 - Reload dm-stage01 and dm-stage02 with RHEL6

Status: RESOLVED WONTFIX

(mozilla.org Graveyard :: Server Operations: Projects, task, P2)

Description

[Dave Miller [:justdave]]

dm-stage01 and dm-stage02 are currently running Fedora Core 7 because RHEL5 wasn't supported by unionfs at the time we originally set them up.

Since then, FC7 has been EOLed, and the unionfs project now supports the RHEL5 kernel.

Reloading those two machines with RHEL5 seems like the right thing to do.  It gets the machines on our RHN subscription so we can track updates more easily, and the kernel is likely to be more stable (how much unionfs will affect that stability is still an open question though).

The services that have currently already migrated from surf to dm-stage01 (cron jobs, mostly) will need to move back to surf long enough to reload the machines.

This depends on the battery replacement for surf's disk array (bug 444886 - closed because it has HP ticket numbers and serial numbers in it), as we need dm-stage02 around as a backup for surf while it's down for the battery replacement.

Comment 1

[Jeremy Orem [:oremj]]

Changing QA Contact.

Comment 2

[Dave Miller [:justdave]]

This is a short-term issue now because dm-stage02 had a catastrophic filesystem failure (which appears to be unrelated to unionfs - it was the root filesystem and not the unionfs mount that was affected, probably more to do with frequent power cycles on the box), and requires a reload anyway just to operate.

Comment 3

[Dave Miller [:justdave]]

dm-stage02 has been reloaded with RHEL5, but hasn't been completely configured again yet. I haven't done dm-stage01 yet.

Comment 4

[Dave Miller [:justdave]]

This isn't blocking anything at the moment with build because they still have surf and their stuff works on surf.  This project is on hold right now pending completion of the phone system upgrade.

Comment 5

[matthew zeier [:mrz]]

found dm-stage01 had crashed some time ago (but after the overheat problem at MPT).  If this is just a reload, can I do that to close this bug out or can we WONTFIX it?

Comment 6

[matthew zeier [:mrz]]

Why bother.  RHEL6 is out.

Comment 7

[Dave Miller [:justdave]]

so reload with RHEL6 instead? :)  (can change the summary for that, don't need a new bug).

Comment 8

[Dave Miller [:justdave]]

Part of this was for better security and logging of what gets uploaded, not just for the virus scan stuff, so it probably still needs to happen eventually.

Comment 9

[Dave Miller [:justdave]]

These two machines are HP DL360 G4s, which are no longer preferred hardware in our current environment.  We'll want to get blades to use for this if this project ever gets resurrected, which makes this now irrelevant.