FAS3140 + NetApp 7.3.2P5 + Continuos Log of: [xxx: iscsi.notice:notice]: ISCSI: New session from initiator "--- ms initiator ---"

[riccardo_paloni]

Hi all.

I have the system described in topic.

Specifically I have 3 windows 2008 r2 cluster with 3 nodes each one (hyper-v role) based on HP Proliant DL580 G5.

I'm using onboard ethernet with hp iscsi license for connectivity to my sanfas, through Microsoft ISCSI initiator.

I'm using also MS MPIO.

Everything works fine but in the syslog of my sanfas i have a lot of notice like that:

[sanfas_xxx: iscsi.notice:notice]: ISCSI: New session from initiator iqn.1991-05.com.microsoft:myhost.mydomain.local at IP addr xxx.xxx.xxx.xxx

I have a trace of notice every 20/30 minutes from all my initiators.

This situation make "real error" in syslog hard to "notice".

I'm sorry for my english, I hope was able to explain my trouble.

Anyone know why is trace this notice?

Riccardo.

[sener29897]

Same issue with FAS8040MC 8.2.1 on Windows 2012 R2

[igor]

I've noticed the same while analyzing an Autosupport message from one of our clients. Exactly every 30 minutes, their Hyper-V server reaches out an refreshes the iSCSI session...

There's no real problem, it's just curious that it should do that so often when compared to Windows, ESX and other servers.

[lafoucrier]

Hi,

I have the same behaviour on my systems (FAS6030 and FAS3140 : 7.3.2P5).

After a long diagnostic with Microsoft and Netapp,  it apears that those messages can be ignored (leading to Netapp and Microsoft) : they are generated on each Hyper-V Hosts by a WMI request orchestrated by SCCM server : Microsoft's answer about this behaviour is : " this is done by Design".

My point of view is that this kind of behaviour could generate useless load, and can lead to freeze VMs, if it occur during an overload of the Storage system.

We have also seen sometime Lun reset messages in syslog comming from Hyper-V hosts, and I suspect that those messages are the consequences of an Iscsi New Session on heavy load, but I do not have any proof.  

If you go further in debug or encounter problems with this behaviour in your environment please let me know.

I continue to investigate with Microsoft until they confirm me that there is no loose of connection and freeze of data access during an "ISCSI new session" (simultaeously on every Hyper-V host in a cluster configuration).

regards,

YL

[riccardo_paloni]

Thank you for your replay.

I'll investingate on same "direction".

I can tell you that problem still present with 7.3.5.1

R.

[DTISTOCKAGE]

Hi,

We just had the same problem. The Default route for the Filer was the management network which is protect by firewal. Excepted for the log, we did not have any iSCSI problem since the Filer always reply in the same network which he had the open session.

Recently, we had to change the default route to the client VLAN for other reason and this solved this problem.

So, we think that the Filer had to send some king of "keep alive session" on the mangement network for iSCSI protocole, which the host can not respond and cause a reset session. When we changed the default route, it was able to reply.

Is it possible?

Thanks

Réal Waite

[dledwards]

Any new info on these messages?

They are ongoing in our Hyper-V environment.  Seems to happen every 20-30 minutes, but if the system is under a heavy load such as a backup, it may be causing problems?

Thanks,

DE