[openstack-dev] [OpenStack][Nova]May be performance issues of connect_volume in Nova
peter.w at huawei.com
Wed Aug 27 02:41:46 UTC 2014
Thanks for your replay.
I want to confirm whether we use “--rescan” for another reason.
If only for connecting volume, it should be a bug. I will report it soon.
From: Joe Gordon [mailto:joe.gordon0 at gmail.com]
Sent: Wednesday, August 27, 2014 1:12 AM
To: OpenStack Development Mailing List (not for usage questions)
Subject: Re: [openstack-dev] [OpenStack][Nova]May be performance issues of connect_volume in Nova
On Tue, Aug 26, 2014 at 5:36 AM, Wang Shen <ws1210 at gmail.com<mailto:ws1210 at gmail.com>> wrote:
I have done some work to test the performance of LUN scanning, use
"iscsiadm" with "--rescan" like what Nova dose. In my test, a host
connected with a lot of LUNs , more than 1000 LUNs. Because "--rescan"
will cause kernel to scan all of the LUNs connected to the host, it
costs several minutes to complete the scanning.
According to "connect_volume" at line 284 in nova.virt.libvirt.volume.py<http://nova.virt.libvirt.volume.py>:
Nova uses "iscsiadm" with "--rescan" to detect new volume, but this
command will scan all of the LUNs, including all the others which
already connected to this host. So if a host has a large number of
LUNs connected to it, the connect_volume will be very slow.
I think connect_volume needn't scan all of the LUNs, only need scan
the LUN specified by connection_info.
Is it necessary to discuss a more efficient way to improve this issues.
It sounds like this is a bug; we use https://bugs.launchpad.net/nova to track bugs so they don't get lost.
====================== Peter.W ======================
OpenStack-dev mailing list
OpenStack-dev at lists.openstack.org<mailto:OpenStack-dev at lists.openstack.org>
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the OpenStack-dev