iSCSI lost connectivity to storage

[Tomas_W]

Hello Everyone,

i am facing the issues, when iSCSI lost connectivity to the storage. Its almost each minute and it seriously driving me crazy. Please see attachment.

Original status:

DELL R820 and R830 + SCv3020. 

Storage is connected to 2 switches via 4 SFP+ cables (DAC)  (each FD on different switch)

Servers are also connected to the same switches.

Switches

Networks (VLANs):

006 – (CORP)

241 – (STORAGE_FD1)

240 – (STORAGE_FD2)

Switches 1 a 2 are connected to each other.

Storage

Storage center FD1

Top Controller        – port 1 -> SW01/25 (STORAGE_FD1)

Bottom Controller – port 1 -> SW01/26 (STORAGE_FD1)

Storage center FD2

Top Controller        – port 2 -> SW02/11 (STORAGE_FD2)

Bottom Controller – port 2 -> SW02/12 (STORAGE_FD2)

Servers

ESXI01 6.20

0 SW02/06 (CORP)

1 SW01/23 (STORAGE_FD1)

4 SW02/10 (STORAGE_FD2)

5 SW01/19 (CORP)

ESXI02 6.21

0 SW02/05 (CORP)

1 SW02/09 (STORAGE_FD2)

4 SW01/21 (STORAGE_FD1)

5 SW01/20 (CORP)

On storage there were 2 volumes and and each server has access only to their volume. No problem with connectivity

-----------------------------------------------------------------------------------------------------------------------------------------------------------

Now i have i got "new" servers -> two DELL R640. Configuration is same as above with difference that earlier it were different servers, with different IP´s.

With new servers, i also created new configuration (scenarios)

In vcenter servers are in cluster. In DELL Storage Manager servers are in also cluster and to this cluster its created one big volume, where are located all machines. So in case host is dead, i can quicly mounted to other host. I go through DELL Best Practises for iSCSI connections. Set for each Fault Domain own switch with one connection which also means no port biding used in configuration. So everything looks ok regarding those best practises, but its still disconnected like each minute and i already do not know, where to look for issue. Cables were completely.

Please help guys!! 🙂

[fabio1975]

Ciao 

Isn't it that you are using Jumbo Frames, that the MTU is set on the vmkernels used for the ISCSI and that the MTU is set on the SWITCH ports? 

It is also not clear to me why in the case of a fault with a host you have to manually connect the volume on the other Host, from what I understand you also have the right vCenter?

[berndweyand · Accepted solution]

do you use the same vmkernel ip address for iscsi on both hosts ?

vmk1 on both hosts have 192.168.241.61, both vmk2 have 192.168.240.61

[Tomas_W]

Hello Fabio,

thanks for answer.

JUMBO frames are not used at all.

I expressed not so clearly. Volume is mounted to both hosts. So both host has access to all VM. It also temporarily coverd us agains issues with some host. Earlier we have to repir that host or completely copy the whole machine from one volume to another (at storage center, server werent in cluster). Its preparation for HA in vcenter. Now we have only essential license which does not have vMotion, HA, etc. We are in the middle of dealing about enterprise license.

[Tomas_W]

Yes

[berndweyand · Accepted solution]

why ? using duplicate ip is never an option

[Tomas_W]

That was it 😄

**bleep** man, thank You. Probably too many IPs on my head 😄

Thank You, Thank You, Thank You