PV MD 3200i - VD not on preferred path

Hi,

I have read almost all posts related with this issue, and I have redistributed (all?) vdisks. Now, I have two VD (100G and 250G), the latter is storing some critcal VMs. and the first one some minor db app. First question:

It's safe to redistribute the vdisk while I/O is running on the block device? I have read on another post and one Dell tech reply that "yes". But I'm concerned, I'm running critical infrastructure here.

Firmware: 07.84.47.60 - Dell md3200i (no external enclosures)

Host os with initiators (open-iscsi) and multipathd(8): Linux. (ubuntu 18.04 AMD64).

I have paths that are OK, but one of theme looks like this: (output from multipath -l from linux):

size=100G features='3 queue_if_no_path pg_init_retries 50' hwhandler='1 rdac' wp=rw
|-+- policy='round-robin 0' prio=0 status=enabled
| |- 12:0:0:1 sdaw 67:0 failed undef unknown
| |- 15:0:0:1 sdau 66:224 failed undef unknown
| |- 16:0:0:1 sdav 66:240 failed undef unknown
| `- 18:0:0:1 sdax 67:16 failed undef unknown
`-+- policy='round-robin 0' prio=0 status=active
|- 10:0:0:1 sdaj 66:48 active undef unknown
|- 11:0:0:1 sdal 66:80 active undef unknown
|- 17:0:0:1 sdai 66:32 active undef unknown
`- 19:0:0:1 sdak 66:64 active undef unknown

 

As I understand this, it has failed over the non-preferred path, which causes the array to complain (the amber light is up, and the message "needs attention" shows up on MDSM).

 

Question #2.

If all other paths on  other vdisks from the same host (4 more) show signs of no problem then what's going on. However. I believe the issue it's not deterministic, as I have seen (before the redistribution) other vdiss were mentioned in the "Needs Attention" window, I then redistributed those (in the maintenane window, all VMs down, etc.) and haven't saw them again as reported on the non-preferred path. But I have this two now.

Yesterday I did a maintenance window and when I redistributed all vidsk directly by going to the Hardware tab, then selecting the controller (this is needed?) then from Storage/Virtual Disk/Advanced -> Redistribute Virtual Disks.

 

Everything was ok, multipathing was working, etc, then like half an hour later 1 path started failing, then another, the two vdisks I  mentioned. As far as I know this "redistribution" is a quick fix (is it?) because yesterday I did just that (but not from right  clicking on the vdisk on MDSM but directly from the Storage menu that I just put in my previus paragraph.) and then the problem appeared again.

 

What I can deduce is that my host conectivity (4 interfaces 1Gbit/s) is OK, I can ping all 8 ip addresses on the SAN (two for each interface, we use 4 subnets to reach controller 0 and controller 1, which are connected to a dell switch), the host is connected to the same switch. Also another host I have have zero problems with iSCSI, using the same switch and the same SAN, same subnets, same number of interfaces. And, since all other paths are showing as correct and are being used (round-robin) I suppose that the problem it's not on my host. MTU is jumbo as it's correctly setup both from the Linux interface and from the Switch in the middle (dell powerconnect 7024) and from the SAN. 

Can it be that I need to recreate those vdisks on the SAN side? 

Also, when doing a wireshark session I can se this message: "LUN Busy" returned from the SAN, I could see this from all interfaces, don't happen all the time but it show up eventually on all of them, is this normal?

 

Thanks I appreciate any insights into this.