[slurm-users] slurmdbd upgrade startup error

[Bjørn-Helge Mevik]

"Tina Fora" <tfora at riseup.net> writes:

> My guess is that there is something in the database that slurmdbd does not
> like. I'm not sure how to debug it further.

You could turn on logging of the actual sql statements that slurmdbd
sends to mysql by adding to the DebugFlags in slurmdbd.conf (see
slurmdbd.conf(5)).  If that doesn't give you all the relevant queries, I
*think* there is a way to get mysql/mariadb to log all queries as well.

>>> slurmdbd: pre-converting job table for cluster
>>> slurmdbd: adding column pack_job_id after id_group in table
>>> "cluster_job_table"
>>> slurmdbd: adding column pack_job_offset after pack_job_id in table
>>> "cluster_job_table"
>>> slurmdbd: adding column mcs_label after kill_requid in table
>>> "cluster_job_table"
>>> slurmdbd: adding column work_dir after wckey in table
>>> "cluster_job_table"
>>> slurmdbd: adding key old_tuple (id_job, id_assoc, time_submit) to table
>>> "cluster_job_table"
>>> slurmdbd: adding key pack_job (pack_job_id) to table "cluster_job_table"
>>> slurmdbd: debug:  Table "cluster_job_table" has changed.  Updating...
>>> slurmdbd: error: mysql_query failed: 1062 Duplicate entry
>>> '1042-1012321342' for key 'id_job'

Alternatively, you could connect directly to the mysql/mariadb database,
look at the table cluster_job_table and see if there really are more
than one row with the same value ('1042-1012321342') for id_job.

If it is so before you try to upgrade, then that probably has to be
fixed first.

If it is not so before the upgrade, but after the (failed) upgrade, then
probably slurmdbd is doing something wrong.  Inspecting the sql queries
it sends might give a hint.

If there is only one row with the id_job value '1042-1012321342', both
before and after the upgrade, then something is really wrong here -- I
believe (but haven't checked) the "1062 Duplicate entry
'1042-1012321342' for key 'id_job'" message comes from mysql/mariadb
itself.

-- 
Bjørn-Helge Mevik, dr. scient,
Department for Research Computing, University of Oslo
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 832 bytes
Desc: not available
URL: <http://lists.schedmd.com/pipermail/slurm-users/attachments/20180815/fe2885ee/attachment.sig>