Nirik, thanks for CC'ing me.

Short version: this issues happens only for bugs (the one mentioned in the ticket too) closed between the Bugzilla query execution and actual run of the Bugzilla's comment.

Longer version: Bugzilla is queried for - version == Fedora 17, status != CLOSED - current query is http://bit.ly/17WmwQY but the EOL script is pretty dumb. It just takes CSV file with the list of bugs as input from the query mentioned and proceeds bugs one by one. This can take up to one day (last run was approximately 15 hours). There's no check if the bug still fulfills criteria. So if the bug is changed between query and script run, there's collision and comment is sent to already closed bug.

Possible solutions:
1. check if bug still fulfills criteria - I'd like to avoid this, not to overload Bugzilla (18000 calls instead of 9000).
2. run in smaller batches of bugs
3. sync with stable push and Bodhi autoclose, this will avoid collision with CLOSED ERRATA bugs, not manual ones...

And I'm open to any other suggestions, ideas. Time to rewrite the script to Python, I can't speak Perl ;-).

Sorry for problems with EOL, but some sort of house keeping in Bugzilla is really needed.

I personally don't know that this is a big deal.

Do we need to solve it? or are the number of bugs affected pretty few?

Usually only a few bugs (I don't have exact numbers) - compared to the amount of bugs to be closed. Just with every EOL warnings sent, I receive these complaints. I can live with that and I always try to to explain it when people asks.

I'd say it's not worth trying to avoid... too much work for too little gain.