[CentOS] Fail2ban & logrotate [was: Update on spam, postfix, fail2ban, centos 6]

[Leonard den Ottolander]

Hello Bob,

On Mon, 2012-06-18 at 10:07 -0400, Bob Hoffman wrote:
> The  debian and redhat issues seem to be worlds apart. I know as I tried 
> all the fixes and found debian fixes a dead end.

I still believe
http://sourceforge.net/tracker/?func=detail&aid=2870788&group_id=121032&atid=689044
is the issue we are seeing.

http://sourceforge.net/tracker/?func=detail&aid=2857096&group_id=121032&atid=689046
might indeed be a different but similar (timing) issue though. The patch
added there is for action.py and indeed not relevant to the jail start
up issues.

> The errors on shutdown are the same as the errors for startup, when not 
> using sleep. I get one for each jail it kills in iptables.

They are similar but not the same (-X instead of -I etc.). I only see
errors for jails for which the iptables rules didn't get applied. The
rules that didn't get added on start up are being removed but not found
which causes the errors on shutdown.

> the sleep command that I presented does prevent the issue on startup.

My bad. The patch you mention indeed does apply. I see why I got
confused:

$ grep def\ __processCmd *
fail2ban-client
server/action.py

I overlooked fail2ban-client and thought this had to be applied to
action.py. I will give that sleep in fail2ban-client a try.

Could you please provide a url to that patch or add it to the bug report
I made? I remember seeing it but forgot to bookmark it.

> I don't know if this is any bug I would submit to redhat, it seemed
> to 
> be fail2ban's issue...

Bugs for EPEL packages have to be reported via Red Hat's bugzilla.
Perfectly valid issue to put in their bug tracker, especially since
there seems to be not much going on upstream. Even confused where
upstream is, there are patches to the SourceForge tree that seem not to
have propagated to the now current(?) GitHub repo.

Regards,
Leonard.

-- 
mount -t life -o ro /dev/dna /genetic/research