Changing path policy requires a reboot?

[geddam]

I got a typical question on my desk, stating

Do we require a reboot of host, if we change the multipath policy on fly...(majorly switching from MRU to RR, or Fixed to RR)

I understand the baselines on pathing for AA and AP arrays...and ESX pathclaaim rules every 5mins, but still looking for a documented note or inputs from experts.

Thanks,,

Ramesh. Geddam,

VCP 3&4, MCTS(Hyper-V), SNIA SCP.

Please award points, if helpful

http://communities.vmware.com/blogs/rameshgeddam

[Troy_Clavell]

no reboot required... We are doing ES4 upgrades and going from fixed to round robin and there are no required reboots.

[geddam]

Troy thanks for input....I too have the same understanding.

The below link, I just came across, where it is stating it requires a reboot. Do you have any comments on this.

http://tech.philipsellers.com/2010/04/07/best-practices-for-vmware-esx4-with-hp-eva-storage/

Thanks,,

Ramesh. Geddam,

VCP 3&4, MCTS(Hyper-V), SNIA SCP.

Please award points, if helpful

http://communities.vmware.com/blogs/rameshgeddam

[Troy_Clavell]

This may be the case for MRU since there is no preferred path, but since we don't use MRU, I can't say for sure.

[chipflyer]

I want to change mine from MRU to RR also and would like to see a definitive answer on this. Surely its been done by someone?

Jeff

[binoche]

new luns added will be managed by new psp after defaultpsp changed;

but exsiting luns were still managed by old psp, please check the results of esxcli nmp device list; now reboot required to let new psp to manage them;

binoche, VMware VCP, Cisco CCNA

[binoche]

ESX pathclaaim rules every 5mins, it is not true; if existing luns have been pathclaimed, it will change by unclaim/reclaim or reboot

usually rescan every 5 mins

binoche, VMware VCP, Cisco CCNA

[geddam]

Not true Binoche.....I think there is mistake here.

http://www.vmware.com/pdf/vsphere4/r40_u1/vsp_40_u1_san_cfg.pdf (Page No: 60), clealy states that, it does path claim every 5mins.

And, I did a test as well my setup, policy changes can be applied on fly, but was still looking for a document note on reboot and we dont have that....I agree with Troy.

Thanks,,

Ramesh. Geddam,

VCP 3&4, MCTS(Hyper-V), SNIA SCP.

Please award points, if helpful

http://communities.vmware.com/blogs/rameshgeddam

[binoche]

thanks for the pdf, its page 60,

"By default, the host performs a periodic path evaluation every 5 minutes causing any unclaimed paths to be

claimed by the appropriate MPP."

only unclaimed paths to be claimed, that means

1, if there are new paths added to luns, these new paths will be claimed by new psp

2, if there are new luns added, these new paths will be claimed by new psp

3, existing luns and old paths do not change

binoche, VMware VCP, Cisco CCNA

[geddam]

So back to question....Does this requires a reboot?>

Thanks,,

Ramesh. Geddam,

VCP 3&4, MCTS(Hyper-V), SNIA SCP.

Please award points, if helpful

http://communities.vmware.com/blogs/rameshgeddam

[binoche]

to let existing luns to be claimed by new psp;

1, reboot or

2, unclaim and then reclaim, for example,

esxcli corestorage claiming unclaim -t location -A vmhba # unclaim

esxcli corestorage claimrule load # new psp rule loaded

esxcli corestorage claimrule run # reclaimed here

my suggestion is to reboot, because the last path of a lun in use maybe not unclaimed and left to be managed by old psp;

and this pdf for your reference http://www.vmware.com/pdf/vsphere4/r40/vsp_40_vcli.pdf

binoche, VMware VCP, Cisco CCNA