iowait high

[bd5hty]

The system architecture is as follows:
XEON E5420 the * 2/24G memory / hard drive SAS300G 15K raid1
Three nodes of the cluster
Each one of the NFS and ISCSI servers

The task when I backup node A within the VM to NFS spent a total of four hours，during IOWAIT value is high, showing 50%, the GUI can not access. Returned to normal until the end of the backup tasks. Is a disk bottleneck or what?

Google Translate

 

[dietmar]

What is the output of

# pveperf

(Please run when there is no load on the system)

 

[bd5hty]

What is the output of

# pveperf

(Please run when there is no load on the system)

root@proxmox-0-161:~# pveperf
CPU BOGOMIPS:      39898.68
REGEX/SECOND:      862230
HD SIZE:           68.41 GB (/dev/mapper/pve-root)
BUFFERED READS:    136.75 MB/sec
AVERAGE SEEK TIME: 4.82 ms
FSYNCS/SECOND:     62.74
DNS EXT:           251.10 ms
DNS INT:           8.24 ms (proxmox.com)

 

[bd5hty]

cpu
Backup mode for GZIP

 

[dietmar]

Your "FSYNCS/SECOND: 62.74" is really low. Do you use write-back cache on the RAID controller?

 

[bd5hty]

Your "FSYNCS/SECOND: 62.74" is really low. Do you use write-back cache on the RAID controller?

Is not clear, on the IDC, the need to verify.This must be enabled?

 

[tom]

write back caches seems not working, check your raid controller/volume settings. you should have at least 1000 here.

 

[bd5hty]

write back caches seems not working, check your raid controller/volume settings. you should have at least 1000 here.

ok,thanks!