This post is more than 5 years old
39 Posts
0
2336
HotSpare Selection criteria..??
Hello everyone,
i've read a primus doc(emc125616) stating that size of failed disk does not matter in the selection of hotspare disk size. it depends on the size of LUN that is created on failed disk.
to test tihis scenario, i've created a RAID5(4+1) on 600GB drives. created one lun of 1 TB and did'nt assigned it to any host. there are 2 hotspares available, one 300GB disk and other 600GB disk. Now i pulled one disk out of so that a hotspares is invoked. i was hoping for the 600GB hotspare to kick in but to my surprise 300GB hotsprae started rebuilding........
do i missreasd the primus document or it works like this way.....waiting for an answer....and to remind u again that LUN(1TB) on that raid is not mapped to any host.
waiting for your kind response...
AnkitMehta
1.4K Posts
0
July 5th, 2012 06:00
Yes, Bus 0 will be scanned. My bad I wrote Bus 4 ..3..2..1...0 (decesending) instead of 0 1 2 3 4 (ascending), multitasking is a... at times.
I will conclude this discussion as below:
On a drive failure, FLARE checks for a Global Hot spare first on the same Bus
If its, unable to find one, as per hot spare algorithm it starts scanning the Bus in a ascending order from Bus 0.
On a Hot Spare selection starting from CX Series it depends on the last LBA (logical blocak address) used on that disk which size to select.
Useful Knowledge Base Solutions for Hot Spares:
emc125616
emc118878
emc151008
emc154563
Useful Document:
EMC CLARiiON Global Hot Spares and Proactive Hot Sparing - Best Practices Planning
Please, mark it answered whichever answered helped you.
Anonymous User
457 Posts
0
July 4th, 2012 03:00
Hi Ali,
Hope you are doing fine…
Please refer the below mentioned document Page-8 , Hope this point(point-2 size:) may answer your question….
http://www.emc.com/collateral/software/white-papers/c1069-clariion-global-hot-spares-ldv.pdf
Regards,
Suman Pinnamaneni
ArsalanAli
39 Posts
0
July 4th, 2012 04:00
"the size of the hotspare selected depends on the size of LUNs bound on the disk being spared, not the raw capacity of the disk"..........does this means that if i used 600GB disk in raid creation, and made one lun of 1TB means each disk is contributing 200GB space....now if one of the disk fails, then the hotspare of 300GB is invoked,cuz the lun is using only 200GB from the failed disk......????
am i conceiving it right..????
AnkitMehta
1.4K Posts
0
July 4th, 2012 07:00
You have got it right!
An appropriate hot spare disk drive is chosen by FLARE, based on the algorithm outlined in the "Global hot spare drive selection algorithm" section of the array. The algorithm determines the size of each available global hot spare and finds the ones large enough to hold the LUNs that are bound on the failed disk in CX Series it depends on the last LBA (logical blocak address) used on that disk. If multiple drives meet the criteria FLARE scans the back-end for the lowest-numbered available hot spare. It will first attempt to select a hot spare on the same bus as the failed drive, and then check alternate buses in ascending order.
Can you check if the Bus location of your failed drive?
This is my understanding:
For Eg. if the failed drive is located on say Bus 4, it will first look for a hot spare on Bus 4 if there is no hot spare bound on it. It will start searching in ascending order i.e. Bus 3, 2, 1, 0 etc as per the alogrithm.
Please, let us know if this was helpful or satisfies your query.
RRR
2 Intern
2 Intern
•
5.7K Posts
0
July 4th, 2012 13:00
I think you explained it very well!
AnkitMehta
1.4K Posts
0
July 4th, 2012 13:00
Thank you, Sir!
ArsalanAli
39 Posts
0
July 4th, 2012 22:00
AnkitM, as for as ascending order is concerned i guess first back-end bus as the drive being hot spared is scanned then BUS 0 will be scanned for the hot spares then BUS1 and so on.....!!!
i guess i got it right too..
ArsalanAli
39 Posts
0
July 5th, 2012 23:00
Thank you very much all guys..
RRR
2 Intern
2 Intern
•
5.7K Posts
0
July 12th, 2012 01:00
Is this for VNX the same as for Clariion?
Do you know the selection criteria for Symm or VMAX as well?
AnkitMehta
1.4K Posts
0
July 12th, 2012 14:00
I am not sure as document mentions only till CX4. However, I believe yes it should be same as they are running on the same OS. (FLARE)
As far as Symmatrix and VMAX is concerned as far as my understanding of Symm is they have whole different concept.
- Dynamic Sparing (DMX )
- Global Sparing (DMAX & VMAX)
You may refer to:
White Paper: Drive Sparing in EMC Symmetrix V-Max Systems — Applied Technology
White Paper: Drive Sparing in EMC Symmetrix DMX-3 and DMX-4 Systems - Applied Technology
If you still have any further queries regarding Hot Sparing on VMAX and Symm I would request you to either Open a Chat Session with Symmetrix Group or You can start a new discussion there!
PS: Congratulations on your 5000 posts!
RRR
2 Intern
2 Intern
•
5.7K Posts
0
July 13th, 2012 01:00
Thanks