Highlighted
tdubb123
3 Zinc

lun tresspass and round robin

Jump to solution

what would be the effect on the esx host and vms when tresspassing a lun on a cx4 when the multipath policy is set to round robin?

how would it differ from if the policy was set at mru or fixed?

thanks

15 Replies

Re: lun tresspass and round robin

Jump to solution

I am going to assume that Failover Mode 4 is being used which enabled ALUA based LUN access and an Active/Active configuration of the storage processors.  I am also assuming 4 paths per LUN/datastore from each ESX host (2 HBAs, 4 array ports).

Let's assume I have VMs on a LUN that is owned by SPA.  All paths will show as Active and the two going to SPA will show Optimized, while the other two going to SPB will be Non-Optimized.  If I issue a trespass command to the LUN, then then the optimized paths will then be attached to the SPB paths and all IOs will process down these paths.

With MRU, a trespass of a LUN will change the Optmized/Non-Optimized setting per path and will cause IO to change to the first available Optmized path.  This is very similar to RR, except in MRU there is not balancing between Optimized paths.

With FIXED, there is a preferred path that is set when the ESX host first talks to the LUN (or manually set), let's say to SPA (the currnet owner).  If I trespass the LUN, the ESX host will continue to send IO to the Non-Optimized path and the storage processor will then send the IO to SPA through internal CMI bus.  There is then an algorithm which will force the trespass of the LUN back to SPA if IOs continue to come in on the Non-Optimized path.

tdubb123
3 Zinc

Re: lun tresspass and round robin

Jump to solution

hello

where do you see the path showing "optimized" ? in vmware, it shows all 8 paths as active (I/O)

I have failover mode 4 enabled as well as useANO=1

0 Kudos
mattcowger1
2 Iron

Re: lun tresspass and round robin

Jump to solution

Because you have useANO enabled, all paths which are 'active' (every path will be marked as active in an ALUA config like yours), everypath will be sent IO (thats what useANO does) even those that are non optimized.

If you were to disable ANO, you'd see what Clint is talking about.

Why do you have that enabled - its generally not a best practice....

0 Kudos
tdubb123
3 Zinc

Re: lun tresspass and round robin

Jump to solution
0 Kudos
mattcowger1
2 Iron

Re: lun tresspass and round robin

Jump to solution

Jason's post is from 2010, and relates to Clariion/Celerra.

Its generally NOT a good idea to have this set - it can cause excessive useage of the CMI and increase latency.

0 Kudos
christopher_ime
4 Ruthenium

Re: lun tresspass and round robin

Jump to solution

I'd like to direct you to a decent comprehensive post regarding ESX integration and block storage from a CX/VNX at the following location:

https://community.emc.com/message/598672#598672

It relates specifically to iSCSI but even though you didn't mention the transport medium, it has many portions that are relevant to both.

tdubb123
3 Zinc

Re: lun tresspass and round robin

Jump to solution

hi

so I change useano to 0. now in vsphere the paths are 8 active but only 4 shows (I/O)

but in esxtop, both hbas still shows traffic active at same time.

is this expected behavior? I thought only 1 hba should be active?

0 Kudos
christopher_ime
4 Ruthenium

Re: lun tresspass and round robin

Jump to solution

The answer is in how you have your HBA's zoned, and the behavior you reported actually validates it was properly architected per best practice.

With the setting now reverted back to the preferred setting (which is also the default), the 4x "Active (I/O)" paths are the optimized paths which correspond to those that are direct paths (via the switch of course) to the current SP owner of the LUN. The other 4x Active (but without I/O) paths are the non-optimized paths or rather those that have a path to the current SP owner but via its peer SP and redirected as necessary through the CMI channel.

Since each HBA have paths to both SPA and SPB per best practice (again the reported behavior validates this), with Round-Robin configured and alternating I/O (1000 by default but never at the same time) down each optimized path, that is why you are seeing activity on each HBA.  If you were to then disable all paths to the current LUN owner, with ALUA configured, you should experience the following:

1) The current Active (I/O) paths should update with status "Disabled" or "Dead" depending on how you are testing or simulating path failures

2) The I/O should transfer to the other 4 Active paths

3) The LUN should not trespass which suggests that the I/O is being redirected by the upper-director via the CMI interface

4) After 128,000 I/O via the non-optimized paths, the system will make the decision to trespass the LUN (implicit trespass) instead of maintaining the non-optimal paths of an extra hop.

christopher_ime
4 Ruthenium

Re: lun tresspass and round robin

Jump to solution

tdubb wrote:

hi

so I change useano to 0. now in vsphere the paths are 8 active but only 4 shows (I/O)

but in esxtop, both hbas still shows traffic active at same time.

is this expected behavior? I thought only 1 hba should be active?

Do you have multiple LUNs?  Remember, the PSP is configured per LUN.

0 Kudos