The cluster is overdue an upgrade to 10.04 from 8.04 (granted, some utility servers are already upgraded).
As such, there are many package upgrades etc that will be dealt with by upgrading the OS
Creating this as a tracking bug for it
Version: unspecified
Severity: enhancement
| Status | Subtype | Assigned | Task | |
|---|---|---|---|---|
| · · · | ||||
| Open | None | T32996 Change $wgCategoryCollation values to appropriate one for each Wikimedia wiki | ||
| Resolved | None | T28388 Upgrade Ghostscript at Commons to 8.71 | ||
| Resolved | tstarling | T48036 Upgrade to ICU 4.8 for WMF | ||
| Resolved | None | T27140 Rendering of malayalam fonts in SVG | ||
| Resolved | mark | T26000 Update rsvg so styling SVG images with CSS works properly on Commons | ||
| Resolved | None | T24708 Update Wikimedia's Mailman install | ||
| Resolved | None | T31905 Getting djvu view components updated | ||
| Resolved | None | T31915 Upgrade the WMF-cluster >= Ubuntu 10.04 | ||
| · · · |
Is this a tracking bug (should depend on the various issues, so them getting fixed makes this one closer to closable) or a bug entry for the upgrade itself (should block them, so that when this one is fixed those can be dealt with)?
(In reply to comment #1)
Is this a tracking bug (should depend on the various issues, so them getting
fixed makes this one closer to closable) or a bug entry for the upgrade itself
(should block them, so that when this one is fixed those can be dealt with)?
I had them as depends originally, but then swapped them over.
These things "should" be fixed (or at least, the upgrade element) when we do the migration of the apaches, image scalers etc etc...
http://rt.wikimedia.org/Ticket/Display.html?id=22 is the RT ticket for this. Ryan Lane is working on this project (updating server binaries) but hasn't said anything since April. CT is also looking into this.
(In reply to comment #3)
http://rt.wikimedia.org/Ticket/Display.html?id=22 is the RT ticket for this.
Ryan Lane is working on this project (updating server binaries) but hasn't said
anything since April. CT is also looking into this.
I think it's a case of too much ops work, not enough ops staff members
Hopefully with the couple of extra new hires, we'll get some more activity soon
The imagescalers can likely be upgraded very soon. Upgrades were tested in the past, but weren't pushed to all nodes. We need to retest with the het deploy changes, then we can upgrade that cluster.
We'll likely start there, then test changes for normal apaches.