Unsolved
This post is more than 5 years old
48 Posts
0
1724
FAST VP, disk showing 100% util, but no IOPS
Hi,
while looking at some analyzer output from my VNX5300, I noticed a 100% utilitzation on my NL-SAS spindle but no iop is going to this disk.
It's a pool with different SAS en NL-SAS in a RAID1/0 setup. Both NL-SAS spindles in the pool are showing the same 100% utilization and 0 IOPS.
Anyone any idea why these spindles are showing a 100% util.
Regards,
Johnn
Kumar_A
727 Posts
0
June 17th, 2011 09:00
Do you have a NAR file that you can share? Can you see the same issue in your archived NAR file? One possible scenario can be that an IO is stuck, since the IO is not getting completed the SP thinks that the disk is busy and reports 100% utilization, but since the IO is not getting completed there are not IOPs. Not saying this is the case, but just giving a possible scenario...
tonydcdi-ymiT1
70 Posts
0
June 17th, 2011 13:00
Is it possible that you are within your relocation window for fast tiering and although that Raid Group may notbe used, it is moving data up and down tiers?
avs
91 Posts
0
June 18th, 2011 01:00
Hello
Are there any LUN binding/Background Verfy processes?
Alex
jjgrinwis
48 Posts
0
June 18th, 2011 10:00
That's the weird thing, nothing is going on this pool lun. Everything is landing on the highest tier and my nl-sas spindles are not used, more then enough space left on highest tier in the pool. No iop no mb's going or coming from these spindles but a constant load of 100%.
Sticky_Bun
7 Posts
0
July 18th, 2011 15:00
Hi
I have exactly the same issue on my VNX5700 with a 24 disk pool configured as RAID 10 which contains only 2TB NL SAS drives (so no tiering here).
2 of the disks are constantly showing as 100% utilization (65 IOPS) even when no load is being put on the LUNs.
Is there a way to determine what is causing this I/O?
Regards
kelleg
4.5K Posts
0
July 19th, 2011 10:00
Quick question - can you look at the disk lights on the front and see if the lights are flashing constantly?
If no flashing, you may want to open a case to have this investigated.
glen
Sticky_Bun
7 Posts
0
July 20th, 2011 04:00
The SP event logs showed it was doing a background verify....hence the I/O when the LUN was not being used.
jps00
392 Posts
0
July 20th, 2011 05:00
A BV like that can be automatically scheduled when a storage processor detects a difference between a group's stripe’s parity and the hard drive’s sector parity.
dynamox
2 Intern
2 Intern
•
20.4K Posts
0
July 24th, 2011 09:00
does VNX run BV when you create a pool LUN or it only applied to RG LUNs ?
Thanks
christopher_ime
2K Posts
0
July 24th, 2011 20:00
Keep in mind, that even with a pool, underneath it is creating private RAID groups and across those, private FLARE LUNs (from which the extents are then allocated). The number of private LUNs created are:
2 * the number of drives in the private RG
For an explanation of how the private RAID Groups are sized based on the amount of disks added and taking in consideration the RAID Type and the recommended multiple, see the comments in the following post:
https://community.emc.com/message/556743
I only wanted to emphasize this as the answer to your question is yes. The BV is not limited to the thin/thick LUNs the user defines on the pool but instead on the private FLARE LUNs across the private RAID Groups. You may have noticed that when you create a pool LUN that the option "No initial verify" doesn't exist unlike when you create a FLARE LUN on a (traditional) RAID Group as it is not relevant as the user doesn't have visibility to the private LUNs.
Basically my only point is that even though it is a pool, underneath exist traditional RG's with traditional/FLARE LUNs and the conditions and behavior for BV applies similarly.
Hope this makes sense.
dynamox
2 Intern
2 Intern
•
20.4K Posts
0
July 24th, 2011 21:00
thank you