On 27.11.20 19:01, Radoslav Bodó wrote:
I have elaborated a bit on your progress and come up with following
patch (attachment) which processes fine on current Bullseye.
Yes, this matches some of my findings.
I also noticed that bacula.BaseFiles has
BaseJobId INTEGER UNSIGNED NOT NULL REFERENCES Job
but bacula.Job does not contain "BaseJobId". Just adding that is more
dangerous since it changes the schema. That has to be vetted by upstream.
And also there is a mismatch in the definition of Client.ClientId being
"INTEGER UNSIGNED" and Job.ClientId and JobHisto.ClientId being only
"INTEGER". The definition all need to match.
Changing this also has to be vetted by upstream, since sometimes in some
tables negative values for Ids have special meaning.
Grüße,
Sven.
