How can i monitor lv activities not using glance?

[jon123_1]

How can i monitor lv activities not using glance?


thanks in advance


jonny

[Rajeev Shukla]

You cant monitor LV activity but you can still monitor the disk using sar.
Do a man and you'll see u can monitor a lot of things and use vmstat for virtual memory statistics.

Cheers
Rajeev

[Michael Tully]

There are no command line porgrams specifically for monitoring LVM performance. You would need to use the native tools such as 'sar' and 'iostat'. Once you identify a disk you can use other tools like 'pvdisplay' to see what logical volumes are affected.

[Brian M Rawlings]

Jonny: It sort of depends on how 'deep' you want to monitor. The earlier suggestions for 'sar' and 'iostat' get great info for deep I/O numbers, but nothing on the health and state of the VG or LV.

A couple of suggestions to monitor the LV/VG state:
1> set up a cron job to run vgdisplay and lvdisplay, redirected to a file. At least do this every morning and have it email the results to you. One of your first tasks in the morning is to check this email, and assure yourself that all is well in your corner of the galaxy.
2> If you send this info to a file, you can then 'grep' it (perhaps using perl, or sed/awk), to look for any file systems that have less than, say, 5% free space. Or, to make sure all your file systems are mounted.
3> You can set up the same sort of scan to parse the /var/adm/syslog/syslog.log file for any references to 'disk' or 'lv' or 'vg' errors. You will have to check log entries from when you've had failures, to get the words to grep for, or maybe be a little creative with your nested pipes to sort through them.
4> You can have cron do this relatively often (every hour, say), if you set your filters up good enough to eliminate false positives. You only want an email if there seems to be a problem, except for that one check first thing every morning (I'd recommend getting that one and reading through it each day, even if you set up other monitoring).
5> You can do the same thing with your backups, if you don't have alerting or reporting functions in the backup SW already doing that for you. Check that all backups completed, watch the syslog.log for tape errors or SCSI errors...

None of this is guaranteed to catch everything, but peace of mind is in short supply, and this may provide a little.

Best Regards, --bmr