IO reduce when remove old data

[DALDON]

Hi all!

I have zm 1.27 with 23 cam, and I want to have more cheap cameras into this server.

So, I use WD40EFRX as my storage drive. It works fine every time expect zm runs a background filter (purge when full).

Please see my iostat output:

Code: Select all

2014-09-19 16:44:01 Device:         rrqm/s   wrqm/s     r/s     w/s    rkB/s    wkB/s avgrq-sz avgqu-sz   await r_await w_await  svctm  %util
2014-09-19 16:44:01 sda               3,10  2165,30   68,30  564,50   300,00 15356,00    49,48    48,05   76,32   19,81   83,16   1,54  97,36

As u can see it uses 97,36 precent of my disk, because of this I can't to login into server via web or ssh.

In a normal time I have:

Code: Select all

2014-09-19 17:02:33 Device:         rrqm/s   wrqm/s     r/s     w/s    rkB/s    wkB/s avgrq-sz avgqu-sz   await r_await w_await  svctm  %util
2014-09-19 17:02:33 sda               0,80   136,30    1,80   71,30    15,60  5725,20   157,07     0,83   11,29   12,67   11,25   2,27  16,56

Can you help me to reduce disk usage when zm purge an old data?

[bb99]

Lower your drive %. I run my TB drive at 80% wih no issues. Try unticking "OPT_FAST_DELETE" to see if that helps.

[DALDON]

Thank you. I have modified : General.pm script for run rm -rf with ionice -c3. It looks like ok, but not enough. It seems that a journal of my ext4 partition eats my I/O with kworker process.

I have checked OPT_FAST_DELETE, but i think it's about to delete an event from a browser only. - That don't help me.

What about your hdd drive? I can't login by ssh when zm purge old data...

I have 23 cams and want more...

[bb99]

By ticking the fast delete option you'll only delete the data base entries and schedule HD removal for when ZMAudit runs next. Unticking forces deletion when the purge when full filter runs. By setting purge when full filter to 80% the drive never gets too full so I can continue to use fast delete. If you're sure it's a drive I/O issue you may want to look into setting up a raid array.