This post is more than 5 years old
14 Posts
0
121807
EqualLogic iSCSI connection limit exceeded - implications?
We have an EqualLogic iSCSI SAN that has exceeded the limits for iSCSI connections. The limit is 1024 but we are at 1612. The thing is that nothing appears to be affected. We can still add LUNs/servers etc. Are we likely to hit problems? Some sort of performance issue? There doesn't appear to be much documentation on this anywhere! We are firmware 5.2.2.
I know what we can do to reduce the connection count, what I want to know is what happens if we don't? Because as far as I can tell nothing bad has happened. Do we ignore it? Panic?
DELL-Joe S
7 Technologist
7 Technologist
•
729 Posts
0
April 30th, 2013 07:00
There isn’t any publically facing information of all the “bad” things that can happen. Basically, you can have anything from dropped connections, to new connections acting unexpectedly. As you can gather, each connection takes a chunk of memory, if you run out of memory for the connection, well… you get the picture.
The support limit currently is 1024 connections per pool (up to 4 pools) or for PS4XXX only groups, 512 per pool (max 2 pools).
You should lower the connection count as soon as you can. If you have a multi-member group, you can explore moving a member to another pool, then move volumes to that other pool to distribute the connections.
-Joe
mrennix
14 Posts
0
May 6th, 2013 23:00
Hi Don,
Yes we are using ESX 5. We have 10 ESX blades, each with 156 iSCSI connections! This is because we have a lot of shared LUNS, average 500GB. Each ESX blade has Dell MEM multipath installed which creates 4 connections per blade to every LUN (there are 39 LUNS).
To get the count down I have done the following:
Editied the /etc/cim/dell/ehcmd.conf file and changed
MemberSessions = 2
to
MemberSessions = 1
then reboot ESX.
This reduces the MPIO connection count from 4 to 2, and the total iSCSI count from 156 to 78 per blade. So over 10 blades we get a reduction of 780 connections, which brings us well under 1024.
My understanding is that this will not weaken redundancy (we have 2 x PS6510X arrays in the pool).
However performance may be slightly affected? Perhaps you could adivse on the downside of this setting.
I will try to get to the latest firmware levels and MPIO levels too.
Many thanks for your answer.
Martin