VMware Cloud Community
syn3rgy
Contributor
Contributor

iSCSI Initiators - Path Status "Not Used"

Hi All,

I currently have 3 ESXi 5 hots, Identical from hardware to vswitch configurations. I am currently setting up the vmkernel ports for the ISCSI initiators using the Dell Equallogic setup script.

It has created the 3 vmkernel ports and each ones is bound to a vmnic, each vmkernal port has one active nic and 2 "not used", when i look at hte 3rd host and look at the iSCSI initiator port bindings, i can see all 3 active paths, however, when i look at host 1 and 3, one path says "not used" and as mentioned above, these 3 host are identical, im just lost as to why the last path is "not used" and yet the host host shows all active which is what i expected to see.

I have stuck some screenshots in a word document attached

Thank you again for your help.

Reply
0 Kudos
13 Replies
mbk4HE
Contributor
Contributor

Hi syn3rgy,

I saw the same issue today and couldn't find any reason. Then I double checked my vmkernel ip settings and found a wrong configured netmask.

With correct ip settings the path switched to online.

Regards

Markus

Reply
0 Kudos
JonesBrandao
Contributor
Contributor

Hello Syn3rgy,

Two questios?

As the image attached the adapter is active in his vmk ISCSI?

You linked to your vmk6 with vmhba?

ex:

esxcli (Name vSwitch) nic add -n vmk2 -d vmhba35

I have helped!

Jones Brandão

Reply
0 Kudos
Boldeman
Contributor
Contributor

I also have a host which currecntly has two un-used ports, netmask and IP are correct as well as the failover settings. It's strange that the behaviour changes after a restart, I have four port groups sometimes two are in use other times three are.

Reply
0 Kudos
syn3rgy
Contributor
Contributor

Hi All,

Sorry I did not reply sooner, been away for a while.

I hvae opened a case with Dell and just awaiting a reply from them.

Yesterday, I changed the patch port on the SAN to my 3rd ESX host which had 3 active paths and now it only has 2, it is really strange.

Once i have an update from Dell, I will let you all know to see if it helps anyone else.

Thanks again for your input guys Smiley Happy

Reply
0 Kudos
Crinos512
Contributor
Contributor

I am also experiencing this issue...

I have 4 ESXi 5.0 hosts with 6 vmk paths each. IP Ranges are valid. Netmasks are the same as active vmks.

iscsi.png

syn3rgy, If you have found the solution to this I would dearly love to hear it.

Reply
0 Kudos
muetstem
Enthusiast
Enthusiast

Equallogic has normaly a active and passive controller in there storage, no active / active configuration. There are Equallogics with 4x1Gbs NICS and with 8x1Gbs NICS but only 1/2 of the NICS are active. So it does not make sense to connect more NICs from your ESXi5 servers to the Equallogic storage.

Reply
0 Kudos
Crinos512
Contributor
Contributor

That is correct. We have a Equallogic PS-6500 and use the Equallogic Multipathing plugin.

That does shed some light on the situation, but I would not say it solves the problem. I would still like to move to an Active/Active configuration.

Oddly, it seems for each volume we have we have an Active/Active/Passive situation where 2/3 of our connectors work, and 1/3 are "Not Used". (See above graphic)

Reply
0 Kudos
Crinos512
Contributor
Contributor

Well, maybe not....

iscsi1.png

Is there possibly a Max number of iSCSI vmks that can be used by the Dell Multipathing Extention Module ?

Reply
0 Kudos
muetstem
Enthusiast
Enthusiast

Not posible, Equallogic only works with active/passive configuration. When you expand with more Equallogic devices you active controllers will

double. This has nothing to do with the Dell Multipathing Extention Module for ESXi storage, but it is the limitation of the controllers on the storage.
But why do you want more NICS? Do you have a problem with throughput? I think that your IOPS with a PS6500 will be more of a problem in the
future. When that happens and you also want more TB you buy a extra PS6500 or a Equallogic with SAS or SSD disks for more IOPS. Your throughput wil grow with a extra PS unit.
Reply
0 Kudos
syn3rgy
Contributor
Contributor

Hi All,

Sorry for not updating this earlier, been a bit manic.

This is the reply I had back from a senior engineer at Dell.

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

The behaviour you are seeing can be expected when you set up the Multipathing Extension Module for ESX.  By default, MEM will only establish 2 sessions per volume slice, & also has options for the maximum number of sessions per volume.  The number of sessions for a given scenario are detailed in Table 6 on page 15 of the user guide.

The preferred method of configuring the MEM is through the esxcli command interface.  An example of this which changes the number of sessions per volume to 3 can be seen on page 18 of the user guide (detailed below):

$ esxcli equallogic param set -n=VolumeSessions -v=3

It is also worth noting that in a single member group, or a group that presents an odd number if physical iSCSI ports, the sessions may not be balanced as "perfectly" as one would expect.  This is because the load balancer is always striving to negotiate a better connection than the one it already has (since connection quality will change depending on the load on a given path).  Sometimes, the only two sessions for a volume will come to port eth0 of a single member group.  This is not a problem since, if the sessions die, & the fabric is redundant, connections into port eth1 should be milliseconds away.

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

My Reply: (Based on my environemnt)

Just to confirm, this behaviour is completely normal.

With regards to the sessions established in terms of Volume sessions, would it be best practice or beneficial to set this to 3? Below shows our info

Name             Value  Max    Min  Description

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

TotalSessions    512    1024   64   Max number of sessions per host.

VolumeSessions   6      12     1    Max number of sessions per volume.

MemberSessions   2      4      1    Max number of sessions per member per volume.

MinAdapterSpeed  1000   10000  10   Minimum adapter speed for iSCSI multipathing.

~ #

Would it be possible to please send me the guide as I have so many guides and yet I cannot find the info you mentioned on those pages (presuming I have the wrong one)

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

Reply from Dell:

I can confirm that the behaviour is normal.  Increasing the number of sessions from the default is beneficial in most cases, but this benefit is more noticeable when there are several hosts connecting, rather than just one.  You may also want to think about aggregating the virtual network links through physical network cards.  In any event, you should benchmark throughput to volumes both before & after any modifications.

The user guide for the MEM can be found here: https://support.equallogic.com/support/download_file.aspx?id=947.  You can also get guides & white papers on VMware integration for various scenarios, as well as other subjects from our support site here: https://support.equallogic.com/support/tech_reports.aspx.  In particular, I recommend that you look at the network performance tuning guidelines.

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

These were my sessions that were established (based on my previouse reply aboce)

sessions.jpg

Reply from Dell with more explination on the above:

I just tried calling you but only got your voicemail.  I wanted to clarify what you are seeing in the screenshot below.  The VolumeSessions count shows the minimum & maximum possible setting for the connections to the volume as a whole, along with the current configured value.  The MemberSessions count shows the same thing for the connections to a volume per member.  This takes into account that a volume may be spread across more than one system in the case of a multi-member group.

In your case, you only have one member in your group, so the MemberSessions count will take precedence.  You should note that the value heading is the current set limit, not the current amount of connections you should expect.  The number of sessions will automatically adjust to the limits you configure as needed.

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

Bit of a lost post but i hope this helps and sheds some light for others that may experience this.

Reply
0 Kudos
Crinos512
Contributor
Contributor

Awesome! I now have 4 paths on each volume and all vmk uplinks are showing as active.

Thank you very much.

As an aside:

( Requires an Equallogic Support Website account )


EqualLogic Multipathing Extension Module for VMware® vSphere ( Version 1.1.0, Date Released: February 2012 )

https://support.equallogic.com/support/download.aspx?id=1484

Recommended PS Series Arrays Firmware ( Version 5.2.1, Date Released:     February 2012 )

https://support.equallogic.com/support/download.aspx?id=1502

Reply
0 Kudos
syn3rgy
Contributor
Contributor

David,

How did you manage to get them all "Active"?

I am not currently running the latest Firmware or MEM 1.1 (running 1.09 i think it is)

Reply
0 Kudos
Crinos512
Contributor
Contributor

I SSH'd into each host and ran:

     esxcli equallogic param set -n=VolumeSessions -v=4

This gave me 4 connections per volume, and between the 5 volumes each host is connected to all 6 vmks are active.

iscsi.png

iscsi1.png

Reply
0 Kudos