Disk Enclosures
1748159 Members
3805 Online
108758 Solutions
New Discussion юеВ

EVA Vdisk Deletion and RSS Issue

 
rccmum
Super Advisor

EVA Vdisk Deletion and RSS Issue

Hi,

Recently we faced a peculier problem in EVA 5000 (2C6D 56x73GB 15K disks) , VCS 3025 , CV 4.0
There are 2DGs , DG1 with 40 and DG2 with 16 disks both with Vraid1.
We tried to delete a Vdisk of 40GB size in DG2 after un-representing from a HP-UX host. After sometime the deletion operation shows " operation failed , try again later". We found the Vdisk was deleted but there
were 8 out of 16 disks got ungrouped!!!! and
This changed the RSS state of the DG2 from "Mirrored" to "None"
We grouped and all those disks back to DG2 and levelling was done. But the RSS state remained "None"

We tried to delete more Vdisks ( size less than 40GB ) thereafter from DG2 but such behaviour was not seen.

We are not sure what could have been went wrong. Is there any way to bring back the RSS state to "Mirrored"?

We are just wondering as why the ungrouping happened at the first place?
and then why didn't the levelling process restore the original RSS mirrored state which was done by EVA while making the DG2?
We do understand that EVA distribute the RSS 6-11 groups and EVA tries to keep them on different disk shelves ( for Vraid1 ). Moreover the levelling ( in case of disk removal or addition ) EVA re-organise the RSS distribute, then why levelling don't regain the original RSS state if you have same number of disks in the DG again?


4 REPLIES 4
NathanK
Regular Advisor

Re: EVA Vdisk Deletion and RSS Issue

#1: I would call HP for support on this.

#2: The DG's don't have RAID protection themselves. So I am understanding that all the vdisks inside the DG's are vraid1. Vraid1 actually mirrors by the stripe to a different drive. Therefore if you lose one drive you actually lose two drives because you have an odd number of disks. I would set your DG's to at least single protection.

Below is the meaning behind each of the RSS states. You may want to do a locate on the RSS and see if this is true for you. If so you may want to trigger a releveling, which changing the failure protection should do.


The redundancy storage set (RSS) has four possible states:

Unknown├в shelves, or possibly the EMU is malfunctioning, preventing the RSS state from being known.

None├в Indicates that all disks are located on the same shelf, and a higher potential for data loss.

Mirrored├в No mirrored pairs in an RSS are on the same shelf.

Parity├в No two members of an RSS are on the same shelf, implying SRC Mirrored.

Unavailable├в The command sent to get state information failed.
If there is no solution then there is no problem
Uwe Zessin
Honored Contributor

Re: EVA Vdisk Deletion and RSS Issue

I agree that this sounds strange - a Vdisk deletion should not drop any physical disk drives from the group!

Unfortunately, VCS does not contain any code to 'fix' a bad RSS state. If you have enough time, you can try your luck with disk group/ungroups, but I am not optimistic.

I would drop the whole group and recreate if that can be done easily or physically move some disk drives.
.
Robert Dobalina
Valued Contributor

Re: EVA Vdisk Deletion and RSS Issue

IT is voodoo on how the EVA distributes disks into a RSS. I would suggest one of two things.

1. Select disks fron CV and figure out yourself which RSS is causing the state to none.

2. Contact HP support. With a SSSU script they can easily tell you which disks on same Enc.

Hp will tell you to shut the system down and move disks as that is supported but I would suggest ungrouping and regrouping with disk movement between enclosures. There is no guarentee that if you ungroup a disk and regroup it that it will go into a new or the same RSS group, but if you have a 6 member RSS and you ungroup a disk that will cause the RSSs to be altered.
rccmum
Super Advisor

Re: EVA Vdisk Deletion and RSS Issue

closed