Unsolved
This post is more than 5 years old
11 Posts
0
2444
Rezoned hosts initiator issues
Hi,
I've recently rezoned a bunch of hosts to VNX5300 (ESXi if that's relevant) and the VNX appears to be a little confused.
Two hosts in particular appear to have pathing inconsistencies and I suspect it's something to do with initiator registration. I'm using Unisphere 1.3.8.1.0119 and the connection status displayed in the Host List is 'Partially Active'.
The HBAs are the same but the zoning may be completely different to how it was done originally. Each host has 2 HBAs, each zoned with VNX via separate switches for multipathing.
I've been doing a bit of googling and from what I can tell it might just be a case of deregistering previous initiators (for these 2 hosts there is 6 initiators listed and I think there should really be 4). 2 Initiators for each host has 'Logged In=No'. If it is set to 'No' does that mean I can deregister them safely?
I've also noticed a Storage Group called '~management' and I'm unsure if any of the hosts/initiators should be in this group. Even the hosts with a status of 'Active' have initiators in this group. Is thje '~management' storage group used for any purpose?
As far as I can tell, all zoning and connections are correct. This is a production environment so don't want to make any rash changes if avoidable.
Any advice will be appreciated.
Cheers,
Tim
ZaphodB
195 Posts
1
December 8th, 2015 07:00
I think you have it right; your hosts have two obsolete initiators each.
I was writing this out, and posted an incomplete reply. On review I found a pretty good write-up that lays out your situation pretty well:
Rebalance VNX storage paths without downtime - FastStorage
timbo475
11 Posts
0
December 9th, 2015 15:00
Thanks Zaphod.
It's kind of a similar situation, except the status of the extra initiators seems to be registered=yes, logged in=no.
It looks like if it's logged in=no, then it probably means I can deregister, but to fix it proeprly, I have to go into 'Storage Groups' Advanced button for the affected hosts and untick the 'phantom' connections. Until I do that, the unregistered initiators still appear in Hosts>Host List. After unticking they disappear form Hosts>Host List and everything *seems* fine.
adhamakady
65 Posts
0
December 10th, 2015 01:00
Hello
If I understand correctly, Unisphere is now showing 6 paths for each host, including 4 logged in (2 to each SP), and 2 logged out (but registered)?
If so, then yes this can be because of the fabric changes. I'd just go ahead and de-register the logged out initiators manually. This wouldn't cause production disruption if all the failover settings are done correctly (failover mode 4 and correct host failover software settings). This is all assuming the zoning is correct of course.
For your question regarding the ~management Storage Group, this is just a group which includes all initiators which are not involved in your user Storage Groups.
Hope this helps.
Post back with any further questions you have.
Adham
ZaphodB
195 Posts
0
December 10th, 2015 06:00
Getting rid of the two paths that are not logged in by de-registering them will clear up your alerts. But the more important part is making sure that the four remaining paths are properly distributed between the HBAs and SPs, and that they are all in the storage group and available for I/O.
kelleg
4.5K Posts
0
December 10th, 2015 12:00
Once you make all the changes it's a good practice to restart the Management Server on each SP. Sometimes the GUI gets confused when changes are made to the initiator registrations.
Go into the SPx_IP_Address/setup page on each SP and click on the Restart Management Server button - again, do one at a time.
Also, for Host connection issues, see KB article 73872 - this lists all KB's about connecting hosts to the array.
glen