FYI there is [1] now (since an hour), which is building atm.
The delta added is [2] to
The code that seems to be related and not working is not using a view, but who knows [3]
self.append('ALTER TABLE %s RENAME TO migration_tmp' % table_name)
(If that is the right code at all, I need to check the package)
I first reproduced it locally with cinder as corey has shown.
it worked with 3.24 and got massive (not just this) errors with 3.26
Note: by enabling cosmic, disco and disco-proposed repos you can up and downgrade to all three.
Also useful [4] on how to run isolated tests of cinder.
Much more tests fail BTW, it seems all of the same class of tests fail. So maybe it is a DB setup issue with the new sqlite.
sqlite 3.24.0-1: works
sqlite 3.26.0-1: fails
sqlite 3.26.0-2: fails
So the next version is not a fix that will unblock us.
Also [5] changelog isn't pointing to anything obvious (I'm sure once we know it later it will be obvious).
We need to isolate the first fail as all others could be follow on issues.
So far it refused be isolated to a reasonable subset of tests.
But I didn't focus on that yet, I'll go back to 3.24 and try to shrink the set.
FYI there is [1] now (since an hour), which is building atm.
The delta added is [2] to
The code that seems to be related and not working is not using a view, but who knows [3] append( 'ALTER TABLE %s RENAME TO migration_tmp' % table_name)
self.
(If that is the right code at all, I need to check the package)
I first reproduced it locally with cinder as corey has shown.
it worked with 3.24 and got massive (not just this) errors with 3.26
Note: by enabling cosmic, disco and disco-proposed repos you can up and downgrade to all three.
Also useful [4] on how to run isolated tests of cinder.
Much more tests fail BTW, it seems all of the same class of tests fail. So maybe it is a DB setup issue with the new sqlite.
sqlite 3.24.0-1: works
sqlite 3.26.0-1: fails
sqlite 3.26.0-2: fails
So the next version is not a fix that will unblock us.
Also [5] changelog isn't pointing to anything obvious (I'm sure once we know it later it will be obvious).
We need to isolate the first fail as all others could be follow on issues.
So far it refused be isolated to a reasonable subset of tests.
But I didn't focus on that yet, I'll go back to 3.24 and try to shrink the set.
[1]: https:/ /launchpad. net/ubuntu/ +source/ sqlite3/ 3.26.0- 2 launchpadlibrar ian.net/ 400530194/ sqlite3_ 3.26.0- 1_3.26. 0-2.diff. gz /gitlab. cern.ch/ cloud-infrastru cture/rally/ commit/ 3995f1fcd03b76f 923637f39cc1b12 f599fa61f6? expanded= 1#21dac577876a8 712e411fa71aa76 8fbd11a87883_ 108_0 /docs.openstack .org/cinder/ latest/ contributor/ testing. html /www.sqlite. org/changes. html
[2]: http://
[3]: https:/
[4]: https:/
[5]: https:/