Page MenuHomePhabricator

make snapshot cron jobs use different configs and mount path than xml/sql dumps
Closed, ResolvedPublic

Description

Right now it's assumed that everything running on these snapshots writes to a single nfs-mounted file system. That's how it's always been. In order to move the xml/sql dumps to use the dumpsdata host filesystem and not move everything at once, we need to permit two mount points and two paths.

Details

Event Timeline

ArielGlenn triaged this task as High priority.

All cron jobs but the wikidata dumps write somewhere on /mnt/data/xmldatadumps/public/other/... except I think for the wikidata cron jobs (@hoo ?) so these should write there with whatever symlink changes need to be made. I know there are already some links for backward compat.

Change 386161 had a related patch set uploaded (by ArielGlenn; owner: ArielGlenn):
[operations/puppet@production] use separate path for public/other datasets

https://gerrit.wikimedia.org/r/386161

All cron jobs but the wikidata dumps write somewhere on /mnt/data/xmldatadumps/public/other/... except I think for the wikidata cron jobs (@hoo ?) so these should write there with whatever symlink changes need to be made. I know there are already some links for backward compat.

Except for the temp dir, the Wikidata entity dumpers write to public/other/wikidata (legacy symlinks) and public/other/wikibase/wikidatawiki (actual data).

The point is that later sometime in the future it will be (say) /mnt/someotherthing/wikidata and /mnt/someotherthing/wikibase/wikidatawiki while /mnt/data/xmldatadumps/public will be some different thing. No symlinks are made or anything into the directory above (public)?

The point is that later sometime in the future it will be (say) /mnt/someotherthing/wikidata and /mnt/someotherthing/wikibase/wikidatawiki while /mnt/data/xmldatadumps/public will be some different thing. No symlinks are made or anything into the directory above (public)?

Only https://dumps.wikimedia.org/wikidatawiki/entities/ and the symlinks inside of other/ mentioned before.

Change 386161 merged by ArielGlenn:
[operations/puppet@production] use separate path for public/other datasets

https://gerrit.wikimedia.org/r/386161

I shoulda closed this when it got merged. Doing so now, as xml/sql dumps are written on a different file system than the cron job misc dumps, this very minute.