Help with 2 node computer storage S/W and XCP 7.5 (XEN)

[jocko]

Apologies,

But I am having a heck of a time getting my trial to work. I don't believe the issue is with StarWind but rather connecting XEN (XCP-NG). I have a 2 node Starwind compute/storage separated deployment following the guide https://www.starwindsoftware.com/techni ... r-Pool.pdf. However it is from 2015 and I am hoping I am just missing something.

The Starwind setup went without a hitch ad is synchronized, due to temporary financial constraints I deviated in that the ISCSI feeds are on a single subnet 192.168.99.20 @ 21 (10G + heartbeat) and the sync channel is 10GB crossover.

I tried running the script to no avail, and finally tried adding the SR manually.

The best I can get is using the below:
xe sr-create type=lvmoiscsi \
shared=true \
name-label="san" \
device-config:target=192.168.99.20,192.168.99.21 \
device-config:targetIQN=* \
device-config:SCSIid=2cf29227e785a37e0
device-config:multiSession="192.168.99.20,3260,iqn.2008-08.com.starwindsoftware:san-filer-1-san|192.168.99.21,3260,iqn.2008-08.com.starwindsoftware:san-filer-2-san" \
device-config:port=3260 \
device-config:SCSIid=2cf29227e785a37e0 \

Upon doing this the storage (SAN) shows the pool master connected but all other nodes as "unplugged", In addition all the local resources (DVD, Local, removable) of the other nodes show unplugged as well.
I can ping and query the ISCSI from the disconnected nodes. Most nodes multipath -ll shows no result, but multipath -v 3 shows
3600508b1001cee7acf40433fa1060548 0:1:0:0 sda 8:0 1 undef ready HP ,
2cf29227e785a37e0 3:0:0:0 sdb 8:16 1 undef ready STARWIND,
2cf29227e785a37e0 4:0:0:0 sdc 8:32 1 undef ready STARWIND,
Sep 26 00:26:49 | Warning: should_multipath() only based on wwids. dev = sda
Sep 26 00:26:49 | wwid 3600508b1001cee7acf40433fa1060548 not in wwids file, skipping sda
Sep 26 00:26:49 | const prioritizer refcount 3
Sep 26 00:26:49 | tur checker refcount 3
Sep 26 00:26:49 | Warning: should_multipath() only based on wwids. dev = sdb
Sep 26 00:26:49 | wwid 2cf29227e785a37e0 not in wwids file, skipping sdb
Sep 26 00:26:49 | const prioritizer refcount 2
Sep 26 00:26:49 | tur checker refcount 2
Sep 26 00:26:49 | Warning: should_multipath() only based on wwids. dev = sdc
Sep 26 00:26:49 | wwid 2cf29227e785a37e0 not in wwids file, skipping sdc
Sep 26 00:26:49 | const prioritizer refcount 1
Sep 26 00:26:49 | tur checker refcount 1
Sep 26 00:26:49 | unloading const prioritizer
Sep 26 00:26:49 | unloading tur checker
Sep 26 00:26:49 | unloading directio checker

Any advice? I really hate asking for help, been doing this for almost 30 years and never asked for help, I see the wwid issue but chasing too far.


Thank you in advance,

[Boris (staff)]

It looks like your XCP-NG servers have some kind of issues with multipathing. I'd recommend you disable and reenable it, both operations followed by a reboot, if possible in your case at all. Additionally, you can check an updated guide at https://www.starwindsoftware.com/resour ... server-7-x to make sure you have done all steps when configuring the environment.
Moreover, from your script I see you use only 2 IPs for iSCSI connections, which in most cases would mean the connection scheme is not correct there (e.g. compute host 1 is connected to a single storage node - 2 or 1 - only. Could you share the diagram of your network setup for me to see how everything is connected there?

[jocko]

Boris,

Thank you for the assistance. Indeed I was attempting to only use 1 10GB connection per Starwind node as a shortcut until my switch arrives next week. I am going to rebuild and do it as you have suggested and see if that celars things up.
Also, thanks for the updated Xen document. I will rebuild and let you know how it goes.

Thanks again,

[Oleg(staff)]

Please let us know about the results.

[jocko]

Just wanted to follow up and let you know the updated guide seems to have solved all my issues. Everything working as expected.
Thank you for such a quick and helpful response!

[Oleg(staff)]

You are welcome!