Performance problem with postfix+maildrop
Hi all, I have a performance problem with a mail server (Dell Power Edge 1750, two xeon 3.06GHz, raid controller and disks: "Vendor: MegaRAID ÊModel: LD 0 RAID5 Ê139G ÊRev: 412W", "Vendor: PE-PV ÊModel: 1x3 SCSI BP ÊRev: 1.1", kernel 2.4), configured with postfix+maildrop. Sometimes, the server doesn't deliver mails anymore, with high load average, and many maildrop processes in "uninterruptible state".
08:53:12 up 146 days, 14:14, Ê1 user, Êload average: 173.08, 173.52, 168 plus 1.33 629 processes: 628 sleeping, 1 running, 0 zombie, 0 stopped CPU states: Ê 2.0% user, Ê 1.9% system, Ê 0.0% nice, Ê96.1% idle Mem: Ê 2068860K total, Ê2059508K used, Ê Ê 9352K free, Ê Ê99464K buffers Swap: Ê 979924K total, Ê Ê17132K used, Ê 962792K free, Ê 774924K cached
From "ps ax":
Hello. I try to execute following bash code thanks to crontab every minute in my computer. -------------------------------CUT HERE------------------------------ #!bin-bash mytime=`-sbin-clock bin-awk '{print $5}'` if ! ping -c 1 google.com; then usr...
terminus:~# ps ax grep maildrop 28737 ? Ê Ê Ê ÊD Ê Ê Ê0:00usr-local-maildrop-bin-maildrop -d 28746 ? Ê Ê Ê ÊD Ê Ê Ê0:00usr-local-maildrop-bin-maildrop -d 28747 ? Ê Ê Ê ÊD Ê Ê Ê0:00usr-local-maildrop-bin-maildrop -d 28756 ? Ê Ê Ê ÊD Ê Ê Ê0:00usr-local-maildrop-bin-maildrop -d
etc... with more than 200 processes with D flag.
From what I can see, it should not be a cpu or memory problem, but a disk i-o related. Anyone can help me?
Thank you,
Alex
Linux groups from Newsgroups