production database server sudden disk usage caused by r3814

[Michael Foley]

As documented in ticket https://rt.admin.canonical.com/Ticket/Display.html?id=99984 r3814 caused a suden 197GB disk usage on the production database.

Filing this ticket retrospectively to track the software bug and fix, and to reference in portal event https://portal.admin.canonical.com/incident/event/655/

[Michael Foley (foli)]

<noodles> foli: so no, there's no bug linked to the MP which fixed the issue at https://code.launchpad.net/~nataliabidart/software-center-agent/fix-store-packages/+merge/317456 that I can see. If you do need to create one, pls link that branch.
<noodles> though I'd be inclined to think the actual issue there is not that bad queries can be run, but that postgres isn't limiting the tmp disk...
<noodles> foli: ^^ which is in https://portal.admin.canonical.com/100011
<foli> yeah, I saw that ticket, and I agree, so I will link that ticket also. but I think a bug in the software can be considered the cause of the alert.