Date: Fri, 19 Feb 2016 19:36:37 GMT Content-Length: 465969478 Last-Modified: Thu, 18 Feb 2016 03:04:32 GMT
Presumably, Last-Modified should be Fri 3:04 GMT. So the last time it ran was 7 pm PT Wednesday, which I thought was after the upgrade, so I thought it survived the upgrade.
| Subject | Repo | Branch | Lines +/- | |
|---|---|---|---|---|
| phabricator: silence stdout of public_task_dump cron | operations/puppet | production | +1 -1 |
| Status | Subtype | Assigned | Task | ||
|---|---|---|---|---|---|
| Resolved | • JAufrecht | T127497 iOS Milestones missing from Phlogiston | |||
| Resolved | • JAufrecht | T125698 Phabricator dump used by Phlogiston should be updated when Phabricator is upgraded | |||
| Resolved | • mmodell | T127524 Phabricator data dump hasn't run automatically since the Feb 17 2016 upgrade. |
Dump script broken by the removal of some columns from maniphest_task. See https://github.com/phacility/phabricator/commit/32225d1dd0f8274ae9323650cbf20ea0571424de
I'm running it again manually after removing those columns from the query. Hopefully soon I will find out if that's the only problem.
Change 272320 had a related patch set uploaded (by 20after4):
Remove 2 obsolete columns from select query fixes T127524
The automated Sunday night run (actually around 7 pm PT, since that's after the dump normally goes) seems to have worked fine and has fresh data up through Sunday. Thanks.
Cron Daemon root@iridium.eqiad.wmnet via wikimedia.org
4:59 AM (10 hours ago)
to root
Traceback (most recent call last):
File "/srv/phab/tools/wmfphablib/phabdb.py", line 1474, in sql_x x.execute(statement, arguments) File "/usr/lib/python2.7/dist-packages/MySQLdb/cursors.py", line 174, in execute self.errorhandler(self, exc, value) File "/usr/lib/python2.7/dist-packages/MySQLdb/connections.py", line 36, in defaulterrorhandler raise errorclass, errorvalue
OperationalError: (1054, "Unknown column 'attached' in 'field list'")
(1054, "Unknown column 'attached' in 'field list'")
rollback!
Traceback (most recent call last):
File "/srv/phab/tools/public_task_dump.py", line 31, in <module> task_phids = [id[1] for id in tasks]
TypeError: 'NoneType' object is not iterable
So that would be the Monday night run I guess. Any ideas?
I guess the patch still needs merging, pulling to the puppetmaster, deploying
Was it run as a live hack, and puppet not disabled on the machine? So, hence, it ran fine manually, then puppet reverted it?
Change 272320 merged by Rush:
Remove 2 obsolete columns from select query fixes T127524
Not sure if that was the final commit to production, and if it was in time to catch the nightly job, but the current dump is still last Sunday's.
Is this blocked by needing code review or something, or is it waiting for a scheduled release? This is increasingly urgent day by day since five teams are using this data. Can I do anything constructive to facilitate this?
Mentioned in SAL [2016-02-27T01:50:57Z] <twentyafterfour> updated phabricator/tools to https://phabricator.wikimedia.org/rPHTO1433a20886a4168d41b81b5ecf5e29687678f2e2 (applying fix for T127524 )
@JAufrecht: the fix is now deployed, hopefully the next automatic dump will work. I can run it manually now if that would help.
A fresh manual dump would help; when will the automatic dump kick in? Tomorrow evening? I think it goes around 1 am UT? Thanks
Nevermind, it looks like the dump was just updated:
-rw-r--r-- 1 root root 450M Feb 27 03:21 phabricator_public.dump
@JAufrecht I'm closing this, reopen if it's still not updating but as far as I can tell it updated automatically on schedule last night.
I see cron output today:
Cron <root@iridium> /srv/phab/tools/public_task_dump.py
NOTICE: rtppl not found!
Is this an issue? If not, please go ahead and close this again.
Thanks for keeping an eye out. It appears that last night's dump ran successfully, because http://phlogiston.wmflabs.org/ shows data as recent as Thu 2016-Mar-03 07:00 PM (PT).
And the dump ran:
Date: Fri, 04 Mar 2016 17:43:31 GMT Content-Type: application/octet-stream Content-Length: 476345151 Last-Modified: Fri, 04 Mar 2016 03:23:00 GMT
We are seeing these again. cron mail with the "NOTICE: rtppl not found!" message. I don't know why that is or if it's an issue yet.
(Also see T172558: Phlogiston data not updated since Aug 2 for recent issues with the dump file.)
Yea, but there it says the reports are fine now as if that NOTICE: rtppl not found doesn't really matter / keep it from creating dumps.
this also appears here: https://www.mediawiki.org/wiki/Phabricator/Trello_to_Phabricator/Research_import#Debug_output
https://github.com/wikimedia/phabricator-tools/blob/master/wmfphablib/rtlib.py
So that is "getting through RT populate tables". and " from rtppl import ppl as users" in rtlib.py.
@chasemp ^ In https://github.com/wikimedia/phabricator-tools/blob/master/wmfphablib/rtlib.py can't find rtppl in "from rtppl import ppl as users" . Do you know where rtppl is?
rtppl was a one-time for migration sensitive file that had emails and associations from rt that was on demand generated. That isn't really a warning so much as a notice that w/o that file all those mappings are null, and throwing it here is a product of wmfphablib/__init__.py importing rtlib.
I'll verify and hopefully remove to clarify
Thank you very much chasemp. The goal is just to silence the cron job one way or another then.
yep, seems like that phab/tools repo is now a mirror in gerrit and needs differential for the change. I don't have arcanist setup at all anymore. We could alternatively silence stdout for the cron :) Commenting out that import or redirecting stdout should work.
Change 372413 had a related patch set uploaded (by Dzahn; owner: Dzahn):
[operations/puppet@production] phabricator: silence public_task_dump cron mails
Change 372413 merged by Dzahn:
[operations/puppet@production] phabricator: silence stdout of public_task_dump cron