EMC Command line find and remedy trespassed luns (this is being run on a EMC VNX 7500)
To list all of the luns that are trespassed on your array run (example output below)
NaviSECCli.exe -h emcvnx7500spa.example.com -user sysadmin -password sysadmin -scope 0 getlun -trespass
LOGICAL UNIT NUMBER 1118
Default Owner: SP B
Current owner: SP A
Ok so now you want to fix the trespass how do you do that?
NaviSECCli.exe -h emcvnx7500spa.example.com -user sysadmin -password sysadmin -scope 0 trespass mine
NaviSECCli.exe -h emcvnx7500spb.example.com -user sysadmin -password sysadmin -scope 0 trespass mine
So what is that last command doing and why am I running it twice?
The trespass command “Trespasses one or more LUNs to the control of an SP” and the “mine” option “Performs whatever trespassing is necessary to take all of the LUNs that have this SP as
a default owner.” Since you might have luns failed over from both SP’s you run it on each sp to have that SP grab control back of its own lun.
Additional note you can run this command from your Celerra or VNX File /nas/sbin/navicli -h emcvnx7500spb.example.com getlun -trespass
I have been working with two new EMC VNX’s and ran into a odd problem and could not find anything about the issue, so I am going to post what I know and see if anyone has any more in-depth knowledge to share about the problem.
After wiring up the fiber ports of the VNX to the Brocade switches I notice that not all of the ports “Logged in” looking a bit farther I noticed that some of the ports were labled as Loop ports.. well that was odd, as the VNX should show as a (F) port and not a Loop (L) port. I dug around EMC’s power link, Brocades forums and could not find anyone else mentioning this problem.. Long story short I ended up setting the prefrences on the individual ports on the switch and forced their speed, and port type to (F) port only. After doing that and the typical warning from Brocade that changing the settings will disrupt traffic (how can you disrupt no traffic??) anyway after forcing the port settings on the switch all worked as expected.
Just in case you are thinking it is a model issue I experienced the same issues with:
EMC VNX 7500 connecting to two Brocade DCX’s 8gb cards and
EMC VNX 57oo connecting to two Brocade SilkWorm 4100’s
If you experience similar issues please post and and maybe get a answer to this issue.
Updating some path types the other day in VMware for a ALUA compliant storage array, came across
Duncan wrote the post very well so as I was running some of the commands I noticed that one of them appeared to be EMC CX specific, so I was wondering what is the proper setting for my EVA that I was running the commands against and for that matter what is the proper setting for other arrays. Here is the command that sets the default Path Selection Policy (PSP) for the array type / luns that you will be adding in the future. You will want to run this command to have all of the hard work you just completed be carried through when you add additional luns, anyway here is the command that Duncan ran:
esxcli nmp satp setdefaultpsp –satp VMW_SATP_ALUA_CX –psp VMW_PSP_RR
This command works great if you are running an ALUA compliant EMC Clariion CX storage array.
but if you are running a different storage array other than an EMC Clariion CX what do you substitute, “VMW_SATP_ALUA_CX”, with?
Well to se a full list of predefined plugins you can run:
esxcli nmp satp listrules
that will give you a list and in my case I should use “VMW_SATP_ALUA” and for that matter it looks like this is the setting you use for all ALUA compliant systems other than the Clariion CX.. So that brings up the question what is special about Clariion ALUA? Maybe a different post….
Taken from http://kb.vmware.com/kb/1008205
Now I have used esxtop from time to time and unless you have a good baseline the numbers can be a bit overwhelming.. Now there is a golden nugget of info in this KB article..
“All arrays perform differently, however DAVG/cmd, KAVG/cmd, and GAVG/cmd should not exceed more than 10 milliseconds (ms) for sustained periods of time.”
The other day I was building a new lun and had the option of raid 5 or raid 6… Hmm.. 6 must be better because it is newer? Not so much.. Is it better? Depends on your criteria of better? More resilient to outages, absolutely. Gives you more storage.. Nope.
So when should you use raid 6? Well with some arrays you have no choice and that is fine, they typically have worked out some of raid 6’s shortcoming, but if you are sitting there with a decision that you normally would have had no problem choosing raid 5, ask your self how quickly are failed drives swapped out? If your answer is less than a day, then you probably will be fine with raid 5.. If your answer is in a term of days, or worse weeks, raid 6 will give you a better chance of keeping your data.
Now please note that for NL SAS and SATA raid 6 is the recommended due to the rebuild time associated, so please take that into consideration, considering the statistics that two drives sitting next to each other have a slightly higher probability of failing than a totally random drive out of your array.
So the other day I was given the task of presenting 32tb contiguous (or as much as possible). No problem I thought, I had a EMC Clariion and a few HP EVA’s at my disposal. All had updated firmware so shouldn’t be a issue any more right? Long story short it was a problem. The two EVA 8400’s would not have been a issue if they had free space. All other EVA’s were 8100’s and they have a max limit of 2 tb. So I moved to the EMC Clariion array, it has a 14tb limit. So here is a little table I came up with:
EMC Clariion and VNX – 14 tb and 16 tb
EMC VMAX – ~59 tb (255 devices at 240 gb)
HP EVA 8100 – 2 tb
HP EVA 8400 – 32 tb (and now with latest firmware release xcs 10 supports snapshots, growing and shrinking luns and CA features)
Netapp – 16 tb
Hitachi 2k product line – 60 tb
One step to remember when doing your P2V process, if able, be sure to run your favorite disk alignment software.
While 2008 servers (and hopefully newer) do not suffer from the alignment issue of a blindly formatted disk 2003 does, and if you are preforming P2V’s one step that can be easily overlooked is a post re-align. While this will cause overhead in P2V process it could give the system the performance boost that will allow your virtualization project a success.