MSA 2012i host link down

[Jeroen H_1]

Since a few weeks I get this in my event log:

Info 2010-02-18 13:26:32
111
A478 Host link up Chan0: 0 Loop IDs
Info 2010-02-18 13:26:32
111
A477 Host link up Chan1: 0 Loop IDs
Warning 2010-02-18 13:26:25
112
A476 Host link down Chan1
Warning 2010-02-18 13:26:25
112
A475 Host link down Chan0

The iSCSI switch gives me no information

Does anybody have an idea how this could be happening?

[Johan Guldmyr]

Hi!

This would usually indicate something wrong with the network connection.

Have you checked the cables or the switch where the MSA is connected?

Can also be caused by having both single and dual path servers connected.

[Jeroen H_1]

Hi!

This would usually indicate something wrong with the network connection.

Have you checked the cables or the switch where the MSA is connected?

Can also be caused by having both single and dual path servers connected.

No I haven't checked the cables. I will replace the cables between the switch and the controllers. ANy recommendation on witch cables to use?

I connect with MPIO but the global host list looks a bit strange:

donkeykong.iq.local
Port A0 & Port B0 & Port B1

spaceinvaders.iq.local
Port A0 & Port B0

iqn.1991-05.com.microsoft:megaman.iq.local
Port A0

megaman.iq.local
Port A0 & Port A1

tombraider.iq.local
Port A0 & Port A1 & Port B0 & Port B1

Thnx for your quick response.

[Jeroen H_1]

I think it's strange that both ports go down simultaneously.

[Johan Guldmyr]

It is pretty strange.

Are those two ports in the same controller? Do they go to the same switch?
Do you only have one controller in this MSA?

[Johan Guldmyr]

Oh, and about the cables.
Maybe you want CAT6 cables but this depends on your environment where the cables run. Maybe you need STP cables if you have a lot of interference.

[Jeroen H_1]

I have a dual controller setup and one 2910al switch.

[Johan Guldmyr]

Might be worth checking out Ethernet-settings on the switch for the ports where you have connected the MSA. Or maybe try other ports in the switch.

[Jeroen H_1]

I wil try a few things. Any idea in which direction I should look. Could it be Flow Control or Jumbo frames?
I am a bit of a newbie in this area. Kind of you to help me out a little bit here.
Any other settings worth looking at?

[Johan Guldmyr]

Well I don't think you should have jumbo frames enabled. Unsure about the flow control - guess the only settings you need to worry about are the ones you can set on the ethernet nic in the web gui of the MSA.

But on the switch, check duplex settings, if there's a fixed speed on the port or if it's auto. I'm not experienced with the 2910al switch.

[Jeroen H_1]

I will have a look. So turning of Jumbo frames is a good idea?
I have every port on 1GB fixed.

[Johan Guldmyr]

Okay, if all ports are 1GB fixed, try to see on the MSA side if you can set the speed in there too.

If not, maybe it's time to check the cables if there could be something wrong there, or the ports in the switch. But it seems a little unlikely as it happens for both ports on only one controller.

Maybe a firmware upgrade could be tried? Have not looked in the release notes but that might be good in either case.

[Jeroen H_1]

I checked and there is actually a new firmware released yesterday. So I will try that one immediately.
Thank you for now. I will post my findings later on.

[Uwe Zessin]

> I have every port on 1GB fixed.

Please configure all ports on the MSA and all switch ports to auto-negotiation.
It is NOT a good idea to set a GbE environment to fixed.

[Jeroen H_1]

Updated the firmware to J210P22 last night and switches everything to Auto-Negotiation. Also disabled Jumbo Frames, just using Flow Control.
Seems to be working ok now accept for my read speeds.

Write speed: 90-100 MB/s (Host to LUN)
Read speed: 8-10 MB/s

Only tested with Explorer copying files.
Don't know how to test is otherwise.

Any idea's on the read speed?

[Jeroen H_1]

I have a 3 Vdisk RAID 5 config with 2 LUN's on each Vdisk.

[Uwe Zessin]

It is been some time that I ran tests on an MSA2012i (G1 model). I've used "iometer" (just find a download location with your favourite internet search machine) in a Windows virtual machine against a 3 or 4-member RAID-5 virtual disk with one volume.

When using large 64 or 128KBytes pure sequential transfers and configuring enough outstanding I/Os I was able to get a little over 100 MegaBytes/sec read and between 60 and 70 MegaByte/sec writes.


For fun I've switched to 512 Byte I/Os. Now I saw 10,000ds of I/Os but only about 5 MegaBytes/sec (as expected ;-)

[Serdar TORAMAN]

I have MSA2312fc and I start to get these messages just after installing the new firmware (M111R06).

I am tired of new problems on each firmware. I assume it is not for solving problems but to create new ones.


Severity Time Event ID Code Message
2011-02-18 22:31:39 B2926 111 Host link up. (port: 2, number of loop ID(s): 2, fabric)
2011-02-18 22:31:38 A554725 111 Host link up. (port: 2, number of loop ID(s): 2, fabric)
2011-02-18 22:28:58 B2925 112 Host link down. (port: 2)
2011-02-18 22:28:58 A554724 112 Host link down. (port: 2)
2011-02-18 22:27:55 B2924 111 Host link up. (port: 2, number of loop ID(s): 1)
2011-02-18 22:27:53 A554723 111 Host link up. (port: 2, number of loop ID(s): 1)
2011-02-18 22:26:21 B2923 112 Host link down. (port: 2)
2011-02-18 22:26:21 A554722 112 Host link down. (port: 2)