[csw-maintainers] Fwd: svn-1.7.5 builds was: svn ingores DESTDIR, was: link error with 1.7.3

[Maciej (Matchek) Bliziński]

2012/6/21 Carsten Grzemba <grzemba at contac-dt.de>:
> Yet another call to programmers: Would anyone care for writing a check
> that would be run against our catalog and report any found problems to
> e.g. the devel@ mailing list? For example, something that builds on
> Carsten's missing libraries heuristics script:
> https://gar.svn.sourceforge.net/svnroot/gar/csw/mgar/gar/v2/lib/python/find_missing_bins.py
> Arming this script with pluggable checks and email output would get
> the job done.
>
>
> is this request still valid? Can you give me more information? I think I can
> help you.

Yup. We currently have no alerting / notification framework, and
problems with catalog go unnoticed until they start cascading. There
was some difference of opinion, where some people argued that checks
should be done synchronously when modifying the catalog (so that you
don't get into the bad state), and I argued that checks should be done
asynchronously (so that they don't slow you down on every single
operation). Either case is possible to implement, and I think that the
async is easier. The reason it's easier is that you don't have to
simulate „what the state of the catalog would be if X was done”, you
just check the current state.

That was about checking. Another issue is alerting / notification. I
haven't researched the open source space for this but I'm sure there
is some project that deals with this, so you can e.g. write a check,
and when it fails, a notification is sent, and you can control how
often the notification will be repeated if the check keeps on failing.
It could be something like Hyperic or Zabbix or Zenoss, or something
like that. In this case we would care only about our own checks for
things like dependency graph cycles, missing dependencies and other
potential problems.

Maciej