Disk Enclosures
1748134 Members
3547 Online
108758 Solutions
New Discussion юеВ

Re: When disk faild in VA7410, I/O throughput was terribly reduce

 
SOLVED
Go to solution
BG Jeong
Advisor

Re: When disk faild in VA7410, I/O throughput was terribly reduce

Thanks Torsten.

I didn't not reconfigure device path yet.
it's too hard to modify path becase DB storage.

I'm frightened that. ^^

Do you teach me step or sample command?

and

if i modify path, how much incease i/o performance? about 10%~20%?

I'm really thank you so much...

ps. attach ioscan log.
Tru64 from Korea
Anthony Cole
New Member

Re: When disk faild in VA7410, I/O throughput was terribly reduce

I have recenty started having the same issues with our VA7410. When the array looses a disk the rebuild slows the array to a crawl and the servers start sending out SCSI aborts.

I am working with HP support to solve the issue, but so far we have not come up with any answers. We do have mixed sizes of disks in our redundancy groups. I wonder how big of an impact this is having on the array.

I would consider replacing the smaller disks with the same size. However I don't think that is an option. Each disk replacment will take a rebuild time of 24 hours. With the performance problems we are seeing, it would make the array completely unusable during that time frame and would take several days or weeks.

I am wondering if this could be firmware related. HP upgraded the firmware not too long before we started having issues.

M/C1 and M/C2 are at version 140
M/C1.B1 and M/C2.B1 are at version 5.0
All LCCs are at HP05
All disks have the latest firmware.
Kevin Lister
Frequent Advisor

Re: When disk faild in VA7410, I/O throughput was terribly reduce

Anthony,

I have had this problem on my arrays and there is only one solution: Buy a better product. The HP VA's are mid-range technology at best. They have some neat features and overall I like the arrays but the performance degradation during rebuilds is a major problem.

I haven't been able to resolved the performance problem but I have been able to reduce the rebuild time by doing the following:

1) Keep the number of drives in the array as low as you can.

2) Run the array in RAID1+0 mode only (not Autoraid or RAID5DP).

3) Keep the array in manual rebuild mode and rebuild it at times when I/O is at a minumum.

4) Before you start the rebuild, remove the failed drive and replace it with a good one. Only do this as long as the array is not giving you the "DO NOT ADD OR REMOVE ANY HARDWARE..." message!

5) HP recommends keeping some empty space in the RG's. Not sure that helps.

Although the performance problem is not resolved on my arrays. I have learned to perform the rebuilds on weekends when the I/O activity on my systems is minimal. I've reduced the array rebuild times from over 24 hours to 8 hours or less by doing the stuff listed above.

I also recommend you let HP perform there diagnostic check on the config of your array as they may find some settings that need to be adjusted. They will send you a report on your array detailing any problems they find and recommended changes to correct the problems.

Wish I could be of more help to you!

kev
Anthony Cole
New Member

Re: When disk faild in VA7410, I/O throughput was terribly reduce

Thanks Kev,

I talked it over with my team and I think that is exactly what we are going to do. We had plans to replace the SAN in the next 3 or 4 months anyway because of the end of support is approaching. Maybe this will get us by until we can get the replacement.

Anthony
Kevin Lister
Frequent Advisor

Re: When disk faild in VA7410, I/O throughput was terribly reduce

Hi Anthony,

You are welcome!

If you can remember to, it would be cool if you could let us know what hardware you decide to go with.

kev