Package: libpam-ssh
Version: 1.92-10
i upgraded several machines yesterday and noticed this morning that
there were dozens (hundreds on one machine) of stuck /USR/SBIN/CRON,
'ssh-agent -s' and 'sh' processes, like this:
root 19497 0.0 0.1 65728 2228 ? S 10:30 0:00 /USR/SBIN/CRON
root 19502 0.0 0.0 11792 436 ? Ss 10:30 0:00 ssh-agent -s
root 19504 0.0 0.0 0 0 ? Zs 10:30 0:00 [sh] <defunct>
i upgraded yesterday at 09:22:48:
# grep pam-ssh /var/log/dpkg.log
2010-03-02 09:22:48 upgrade libpam-ssh 1.92-8 1.92-10
2010-03-02 09:22:48 status half-configured libpam-ssh 1.92-8
2010-03-02 09:22:49 status unpacked libpam-ssh 1.92-8
2010-03-02 09:22:49 status half-installed libpam-ssh 1.92-8
2010-03-02 09:22:49 status half-installed libpam-ssh 1.92-8
2010-03-02 09:22:49 status half-installed libpam-ssh 1.92-8
2010-03-02 09:22:49 status unpacked libpam-ssh 1.92-10
2010-03-02 09:22:49 status unpacked libpam-ssh 1.92-10
2010-03-02 09:24:42 configure libpam-ssh 1.92-10 1.92-10
2010-03-02 09:24:42 status unpacked libpam-ssh 1.92-10
2010-03-02 09:24:42 status half-configured libpam-ssh 1.92-10
2010-03-02 09:24:43 status installed libpam-ssh 1.92-10
and the breakage started immediately:
# grep pam_ssh /var/log/cron.log
Mar 2 09:25:01 indra CRON[3631]: pam_ssh: 3634: No such process
Mar 2 10:20:01 indra CRON[12007]: pam_ssh: 12010: No such process
Mar 2 10:20:02 indra CRON[12015]: pam_ssh: 12010: No such process
Mar 2 11:20:01 indra CRON[21240]: pam_ssh: 21243: No such process
Mar 2 11:20:02 indra CRON[21248]: pam_ssh: 21243: No such process
Mar 2 12:20:01 indra CRON[30374]: pam_ssh: 30377: No such process
Mar 2 12:20:02 indra CRON[30382]: pam_ssh: 30377: No such process
...
Mar 3 09:20:02 indra CRON[32565]: pam_ssh: 32560: No such process
Mar 3 10:20:01 indra CRON[9545]: pam_ssh: 9548: No such process
Mar 3 10:20:02 indra CRON[9553]: pam_ssh: 9548: No such process
# grep pam_ssh /var/log/cron.log | wc -l
55
it only stopped when i ran pam-auth-update and disabled "Silent SSH-key
single sign-on"
this happened on all 5 machines that i upgraded yesterday. my guess is
that pam's common-session is trying to start ssh-agent when the cron job
starts up, but is unable to do so for some jobs
note that while most of the jobs listed above are at 20 minutes past the
hour, not all of them are....and i don't actually have any cron jobs
that ONLY run at 20 past the hour (but i do have a few that run every 5
minutes)
on this machine (indra), most of the errors do occur at 20 past the hour:
indra:~# grep pam_ssh /var/log/cron.log | awk '{print $3}' | awk -F: '{print
$2}' | sort | uniq -c | sort -n
1 05
1 25
3 00
50 20
but on another machine, they don't:
kali:~# grep pam_ssh /var/log/cron.log | awk '{print $3}' | awk -F: '{print
$2}' | sort | uniq -c | sort -n
2 00
2 27
10 40
14 38
16 20
18 10
50 05
52 35
54 55
which is a long-winded way of saying that it doesn't seem to be tied to
any one cron job...rather it seems that it can happen on any cron job
but doesn't happen every time that job runs (but those that run more
frequently are, of course, going to trigger it more often).
craig
ps: not everyone will have cron.log enabled in their syslog/rsyslog
config. if you want to look for this in your own logs, these log entries
can also be found in /var/log/syslog.
--
craig sanders <c...@taz.net.au>
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
