Koozali.org: home of the SME Server
Obsolete Releases => SME Server 9.x => Topic started by: mophilly on July 24, 2019, 01:14:00 AM
-
I have been chasing a performance problem. Not much joy in resolving it but I am learning a lot (I hope).
atop reports:
LVM | main-root | busy 101% | read 0 | write 404 | KiB/r 0 | KiB/w 3 | MBr/s 0.0 | MBw/s 0.2 | avio 24.8 ms |
MDD | md1 | busy 0% | read 0 | write 406 | KiB/r 0 | KiB/w 3 | MBr/s 0.0 | MBw/s 0.2 | avio 0.00 ms |
DSK | sdb | busy 100% | read 0 | write 75 | KiB/r 0 | KiB/w 21 | MBr/s 0.0 | MBw/s 0.2 | avio 133 ms |
DSK | sda | busy 1% | read 0 | write 74 | KiB/r 0 | KiB/w 21 | MBr/s 0.0 | MBw/s 0.2 | avio 0.92 ms |
I have not encountered this before. Average process cpu is 3%, I see a lot of processes for httpd, php-cgi, qpsmtpd-forkser, fail2ban and postmaster.
I am beginning to think the SSD's are trying to tell me to replace them. However, the short smart test looks OK.
What else should I be looking at?
-
This is a follow up for anyone who may be interested. After a lot of investigation of applications and the configuration, and correcting two problems we found in the process, we noticed a high error rate on one drive in the RAID array. At this point, the working theory is we need to replace the drives in the array; both are four years old. First, replace the one we feel is faulty. After the array is rebuilt and sync'd, replace the remaining one and let it sync.
-
And the final follow up. We followed the instructions for upgrading the disks on the page https://wiki.contribs.org/Raid (https://wiki.contribs.org/Raid). All went well and the move from 240GB SSDs to 500GB SSDs was completed in about 2.5 hours, start to finish.