System P4 2.4GHz/512M/1 GB swap/ RedHat 9.0 + updates

I ran the command-line version of Seti for the first time last week.
Since then [though it may just be a coincidence] I have noticed that even after disabling and now deleting Seti:

a) Sendmail and Procmail are high up in the 'top' listing, with Procmail taking 450M of memory or more ! Needless to say the system grinds to a halt eventually - windows and icons are slowly painted row-by-row until you can't even log out or switch to another console Ctrl-Alt-F2 to shut down.

b) The directory /var/spool/clientmqueue gradually fills up with files called dxxxxxxxxxxx where x...x is an alphabetic string, each up to 2 GB in size.

c) checking the 'mail' mailbox I find some large messages apparently sent by cron. No user has a cron entry.

If I disable Sendmail from starting as a service at boot time, I don't get this problem - however I can't consider this as a proper fix !

As soon as I restart Sendmail, the decline begins, and I have to reset or power off if it's left too long.

Something has become trashed somewhere, and I'm blaming Seti at the moment !

Ive used seti on and off for years and never seen or heard of this. There is a remote exploit that effects sendmail that sounds alot like what your describing. What platform and version of sendmail are you running?


See http://www.cert.org/advisories/CA-2003-25.html

Platform is P4, as above, version is 8.12.8, so I guess I shouldn't rely on RedHat update to keep track of security fixes, since that link states that the fix is in 8.12.10.

Even so, I don't think this is the problem.

More likely that something got screwed up somehow (OK not necessarily Seti - it was just a case of "post hoc ergo propter hoc" so I probably jumped to the wrong conclusion).

I was hoping that someone would recognise the symptoms and say "oh, you've got XXX misconfigured and that's making cron report lots of spurious mail messages to root" - or something similar.

Will run without sendmail for now - or back up and reinstall RH if things get any worse :-(

dont know if you have fixed it but we have had the very same problem

all the inodes on the disk had been used by the creation of all these files in the /var/spool/clientmqueue

the end users have created some cron jobs which when fail are trying to send emails to the root user via the cron service but sendmail is not setup correctly on our server.

to fix the problem (Redhat rel3) we edited the file /etc/crontab and change the enrty MAILTO=root --> MAILTO="". this corrected the problem but prevents all cronjobs from sending email (also set "chkconfig sendmail off" as sendmail is not required on our server")

hope this helps (I know the post is old)

regards

sam

Found some possible mention of it here: http://www.onlamp.com/pub/a/bsd/2002.../securing.html

Maybe I inadvertantly had the firewall enabled on my desktop PC ?

Anyway, I no longer work at that site now !