[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Process check always running restart script
From: |
Brian Youngstrom |
Subject: |
Re: Process check always running restart script |
Date: |
Wed, 21 May 2003 16:21:48 -0700 |
I feel really stupid. I had a patch in my spec file that increased the
size of bufsize to support a large number of classes (one for every rpm
on the system). I rebuilt the rpm after upping the rev, but not
removing the patch. Removing this patch, all is well.
Before I found this, I created two configs: one with problems, the other
without. I found in the -d2 run of the problem configs' '/bin/ps auwx'
that it wasn't returning data. The working configs' run of ps was
returning data.
Sorry to have bothered everyone :)
-Brian
On Wed, May 21, 2003 at 10:50:22AM +0200, Mark Burgess wrote:
> On 20 May, Brian Youngstrom wrote:
> > I've been looking at this some more. Its still happening in 2.0.7 btw.
> >
> > If I run 'cfagent -vqn', it correctly sees all processes running and
> > determines that nothing needs to be done:
> >
> > cfengine:frida: Running process command /bin/ps auxw
> > Defining classes
> > DoSignals(sshd)
> > cfengine:frida: Matches found for sshd - no restart sequence
> > Defining classes
> > DoSignals(portmap)
> > cfengine:frida: Matches found for portmap - no restart sequence
> >
> > When I drop the -n, the restart script is run, even though all
> > processes are running.
> >
> > Any ideas as to what is happening? I'd love to try any suggestions.
> > I'm currently still at 2.0.5 because of this bug.
> >
> > Thanks,
> > -Brian
> >
> >
> > On Tue, Apr 29, 2003 at 11:24:13AM -0700, Brian Youngstrom wrote:
> >> I'm having some problems with some of my hosts after upgrading to
> >> v2.0.6. All machines are Redhat 7.3 w/ local kernel.
> >>
> >> In my input file, I have
> >>
> >> processes:
> >> "syslogd" restart "/etc/rc.d/init.d/syslog restart"
> >> useshell=false
> >> inform=true
> >>
> >> "ntpd" restart "/etc/init.d/ntpd restart"
> >> useshell=false
> >>
> >> and so on. This has worked perfectly on all machines up through
> >> v2.0.5. After upgrading, some of my machines are always running the
> >> restart script.
> >>
> >> Snippets of cfagent -d2:
> >> Actionsequence item processes
> >> cfengine:frida: Running process command /bin/ps auxw
> >> cfpopen(/bin/ps auxw)
> >> Appending [USER PID %CPU %MEM VSZ RSS TTY STAT START
> >> TIME COMMAND] Appending [root 1 0.0 0.0 1368 432 ?
> >> S Apr07 0:04 init] Appending [root 2 0.0 0.0 0
> >> 0 ? SW Apr07 0:01 [keventd]] Appending [root 3
> >> 0.0 0.0 0 0 ? SWN Apr07 0:04 [ksoftirqd_CPU0]]
> >> [...]
> >> Appending [root 3883 0.0 0.1 1656 652 ? S 10:11
> >> 0:00 syslogd -m 0] Appending [root 3896 0.0 0.0 1364 440 ?
> >> S 10:11 0:00 klogd -x] Appending [root 4001 0.0 0.1
> >> 1468 544 ? S 10:11 0:00 /usr/sbin/smartd] Appending [ntp
> >> 4541 0.0 0.3 1884 1876 ? SL 10:15 0:00 [ntpd]]
> >> [...]
> >> Regex root 4684 0.0 0.1 1656 652 ? S 11:01 0:00
> >> syslogd -m 0 matched syslogd ProcessFilter(root ...)
> >> Matched proc[root 4684 0.0 0.1 1656 652 ? S 11:01
> >> 0:00 syslogd -m 0] Found matching pid 4684
> >> Defining classes
> >> DoSignals(syslogd)
> >> cfengine:frida: Matches found for syslogd - no restart sequence
> >> [...]
> >> Regex ntp 4541 0.0 0.3 1884 1876 ? SL 10:15 0:00
> >> [ntpd] matched ntpd ProcessFilter(ntp ...)
> >> Matched proc[ntp 4541 0.0 0.3 1884 1876 ? SL 10:15
> >> 0:00 [ntpd]] Found matching pid 4541
> >> Regex ntp 4719 0.0 0.3 1888 1880 ? SL 11:01 0:00
> >> [ntpd] matched ntpd ProcessFilter(ntp ...)
> >> Matched proc[ntp 4719 0.0 0.3 1888 1880 ? SL 11:01
> >> 0:00 [ntpd]] Found matching pid 4719
> >> Defining classes
> >> DoSignals(ntpd)
> >> cfengine:frida: Matches found for ntpd - no restart sequence
> >>
> >>
> >> so it looks like cfengine sees the processes there, recognizes that
> >> nothing should happen, but runs the restart script anyway.
> >>
> >> Any idea why this would be happening?
> >>
> >> Thanks,
> >> -Brian
> >>
>
>
> Can you provide me with a simple example script that always has the
> problem? I cannot seem to see the problem here.
>
> M
>
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> Work: +47 22453272 Email: Mark.Burgess@iu.hio.no
> Fax : +47 22453205 WWW : http://www.iu.hio.no/~mark
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
--
Brian Youngstrom
byoung@cs.washington.edu
University of Washington
Computer Science & Engineering