Bug#1062979: tango-db: the upgraded version of tango-db is not compatible with maria-db from bookworm

Thomas Braun thomas.braun at byte-physics.de
Thu Nov 28 10:31:38 GMT 2024


Am Freitag, dem 08.11.2024 um 10:46 -0300 schrieb Santiago Ruano
Rincón:

[resending to bug, as the original mail only went to Santiago]

> Control: found -1 tango/9.3.4+dfsg1-2
> 
> El 27/10/24 a las 12:28, Thomas Braun escribió:
> > > Santiago Ruano Rincón <santiagorr at riseup.net> hat am 25.10.2024
> > > 20:15 CEST geschrieben:
> > > Sorry for the noise. I can confirm there is an issue in bookworm,
> > > and it
> > > is present even without upgrading from bullseye. Which makes
> > > sense with
> > > https://gitlab.com/tango-controls/TangoDatabase/-/issues/69
> > > 
> > > I will propose a bookworm update soon.
> > 
> > Yes, this is indeed an issue on bookworm as well. But the problem
> > only occurs if you use the table and not when you create them.
> 
> I believe that the above matches my understanding. And that is why I
> am
> surprised of not being able to find (reproduce) the issue described
> by
> Frédéric when using the table (created in bullseye) to add the
> starter
> server in bookworm. I gave the steps described by Frédéric another
> try,
> but I still cannot reproduce it. More info is welcome.
> 
> Thomas, other than registering a Starter in the database, do you have
> examples of other use cases that use the table and that would be
> affected by the bug?

Defining a device server is not enough to trigger the bug. You need to
write into the property history tables.

You can do that by adding a property, something like

tango_admin --add-property sys/tg_test/1 key value

triggers the bug here

2024-11-10T17:13:30,966685+0100 DEBUG (DataBase.cpp:6678)
sys/database/2 DataBase::PutDeviceProperty(): sql_query INSERT INTO
property_device_hist SET
device='sys/tg_test/1',id='1',name='key',count='1',value='value'
2024-11-10T17:13:30,966716+0100 WARN (DataBaseUtils.cpp:463)
sys/database/2 DataBase::db_put_device_property failed to query TANGO
database:
2024-11-10T17:13:30,966721+0100 WARN (DataBaseUtils.cpp:464)
sys/database/2   query = INSERT INTO property_device_hist SET
device='sys/tg_test/1',id='1',name='key',count='1',value='value'
2024-11-10T17:13:30,966725+0100 WARN (DataBaseUtils.cpp:465)
sys/database/2  (SQL error=Field 'date' doesn't have a default value)

> > Based on Santiago's MR
> > https://gitlab.com/tango-controls/TangoDatabase/-/merge_requests/82
> >  I've created a new one with a test and an upgrade script, see
> > https://gitlab.com/tango-controls/TangoDatabase/-/merge_requests/95
> > .
> > 
> > Please test.
> 
> I am including your patch into the 10.0 version I am going to upload
> to
> unstable, dropping my patch that was included in 9.4.2+dfsg1-1.
> According to the CI and the autopkgtest:
> https://salsa.debian.org/science-team/tango/-/jobs/6552116
> the Starter is successfully registered (and started).
> 
> I will backport your patch to bookworm afterwards. Sorry for the
> delay,
> I was planing to work on bookworm earlier, but I should have
> mentioned
> that it is good to expose the changes in unstable first. And for
> including them into unstable, I needed to fix some issues I found
> (and
> had overlooked) with opentelemetry-cpp so I can upload tango 10.0.

Yes, definitly let it ripen a bit in unstable.

Thomas



More information about the debian-science-maintainers mailing list