On Sun, 24 Apr 2011 18:54:03 -0300
Rafael Gomes <rafaelgomes(a)techfree.com.br> wrote:
Guys,
I am taking a look in noc01 Nagios Web interface and I am seeing a
alert with 6 hours without any comments, just a "no notification" tag.
Should we remove that check or make some change in server (hosted01)?
This would be a great one for someone to investigate more. ;)
The problem I think is that gitweb-caching or gitweb (it's unclear
which one hosted01 is using, possibly both), is generating Zombie
processes when it finishes:
apache 10034 0.0 0.0 0 0 ? Z 23:10 0:00 [git-cat-file]
<defunct>
apache 18628 0.0 0.0 0 0 ? Z 23:28 0:00 [git-rev-parse]
<defunct>
apache 18638 0.0 0.0 0 0 ? Z 23:28 0:00 [git-rev-parse]
<defunct>
apache 20580 0.0 0.0 0 0 ? Z 23:30 0:00 [git-ls-tree]
<defunct>
So, what needs to happen here is someone needs to look and determine:
- Are we using gitweb-caching or gitweb ?
- Which one is causing these Z processes?
- Can we switch to the other one or fix things so there are no more
Zombie processes?
So, I think the check is fine and valid and showing us a real problem.
We need to investigate why it's happening and how to fix it. ;)
kevin