poimport exhausts disk space on db server
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Launchpad itself |
Fix Released
|
Critical
|
Данило Шеган |
Bug Description
In under 2 hours, poimport managed to cause the PostgreSQL database to chew up over 150GB of disk space. The process was killed before it could exhaust one of the partitions entirely.
I have the start of the query that was currently being executed:
SELECT
POMsgId.msgid AS msgid,
POMsgID_
context,
date_reviewed,
is_current,
is_imported,
pt0.translation AS translation0,
FROM POTMsgSet
JOIN TranslationTemp
Translati
JOIN POTemplate ON
POTemplat
JOIN POFile ON
POFile.
POFile.
JOIN TranslationMessage ON
POTMsgSet
(
POFile.
I don't have the rest of the query, so I can't investigate further. The disk space will have been chewed up by temporary tables, either explicitly created by poimport or implicitly as part of this query so hold intermediary results. I don't know if the query is broken, if there is something about this particular import that caused too many results to be generated, or if a really bad query plan was chosen by chance.
Related branches
- Brad Crittenden (community): Approve
- Diff: None lines
- Henning Eggers (community): Approve
- Diff: None lines
Changed in rosetta: | |
status: | Triaged → Fix Released |
security vulnerability: | yes → no |
visibility: | private → public |
Changed in rosetta: | |
status: | Fix Committed → Fix Released |
First step is to try to engineer the actual query being executed.