Skip to content

Changing value of disk queue_depth parameter

It has been noted installing VIO clients with virtual disk adapters and MPIO that the disk’s original queue_depth values are small, too small indeed.

lsattr -El hdisk## | grep queue_depth
queue_depth    3       Queue DEPTH            True

As this quickly becomes an issue as the systems will “look” slow we have to respond changing the queues to a more reasonable value. If there are only a few disks the change is easy. But as the number of disks grows the amount of effort required to do the job grows too. Bellow, you will see how this change can be implemented across all disks except the ones that belong to the rootvg. By the way, for this change to be effective the volume groups containing these disks should be varried off which implies that their file systems will have to be un-mounted. So, with this two conditions satisfied, lets execute straight from the command line:

for pv in `lspv -v "hdisk0|hdisk1" | awk -F '{print $1}'`
>chdev -l pv$ -P -a queue_depth=32

The > character will by automatically added by the shell – do not enter this symbol! Now, vary on the volume groups, mount their file systems and proceed to the next task on your list.

To validate the change, type in:

for pv in `lspv -v "hdisk0|hdisk1" | awk -F '{print $1}'`
>echo $pv
> lsattr -El $pv | grep queue

Posted in Real life AIX.

Tagged with , , , , , , , .

2 Responses

Stay in touch with the conversation, subscribe to the RSS feed for comments on this post.

  1. Randy says

    Can you change the queue depth with the -P switch while the VG’s are online and have the changes take effect with teh next reboot?
    chdev -l hdisk10 -P -a queue_depth=32

  2. MarkD:-) says

    Yes, you can.

Some HTML is OK

or, reply to this post via trackback.

WordPress Anti Spam by WP-SpamShield

Copyright © 2016 - 2017 Waldemar Mark Duszyk. All Rights Reserved. Created by Blog Copyright.