Disk I/O is overloaded

[hgomez]

Hi to all,

We used Zabbix 1.8.14 previously and are now monitoring our ESXi infrastructure using Zabbix 2.0.2 and Zabbix Proxy 2.02 for remote site and Zabbix Java Gateway 2.0.2 for JMX.

I get many errors about "Disk I/O is overloaded" :

Trigger: Disk I/O is overloaded on drw-hudson-sup.mycorp.com
Trigger status: PROBLEM
Trigger severity: Warning
Trigger URL:

Item values:

1. CPU iowait time (drw-hudson-sup.mycorp.com:system.cpu.util[,iowait]): 25.29 %
2. *UNKNOWN* (*UNKNOWN*:*UNKNOWN*): *UNKNOWN*
3. *UNKNOWN* (*UNKNOWN*:*UNKNOWN*): *UNKNOWN*

And some time later, problem seems to be fixed :

Trigger: Disk I/O is overloaded on drw-hudson-sup.mycorp.com
Trigger status: OK
Trigger severity: Warning
Trigger URL:

Item values:

1. CPU iowait time (drw-hudson-sup.mycorp.com:system.cpu.util[,iowait]): 0.27 %
2. *UNKNOWN* (*UNKNOWN*:*UNKNOWN*): *UNKNOWN*
3. *UNKNOWN* (*UNKNOWN*:*UNKNOWN*): *UNKNOWN*

Latest data graph show there is such high cpu.iowait values

Did others get same behaviour ?

Some systems on our infra are still using Zabbix Agent 1.8.x and we see that some metrics, like free disk space reported as MB instead of GB, could be wrong.

What about cpu.iowait provided by Agents 1.8.x vs Agents 2.0 ?

Thanks all for your time

[Colttt]

you have an IO problem!! you have a very high read/write work.. what do you use?? SAS/SATA/PATA/SCSI-HDD?? maybe a HDD is down?? which RAID-level do youn use?

[hgomez]

Our physical machines are using SAS 10k RAID 5 disks and we're using ESXi 4

Some of our VMs reports such iowait problems.

[quest]

Did you or someone else resolve that problem? - I have the same one, my zabbixserver is a VM (HyperV).

[Colttt]

if you have high IO you might to use a better storage..

or try something like that:

echo noob > /sys/block/sda/queue/scheduler

(only use it if you use virtualizing)

in our enviroment it helps a lot (look at the screenshots)

[[email protected]]

Zabbix VM (Opensuse) echo noob > /sys/block/sda/queue/scheduler

Hello,

Sorry, am a newbie, I tried your solution of:
echo noob > /sys/block/sda/queue/scheduler
but get error "-bash: echo: write error: Invalid argument".... am using Zabbix 2.4 Appliance which runs on a VM on WIndows 2012 Hyper-v for test purposes... please could you assist?

Thanks,
Stefan

[okelet]

It should be:

echo noop > /sys/block/sda/queue/scheduler

More info https://lonesysadmin.net/2013/12/06/...tual-machines/