VA7400 Hot spare/redundancy

[Gregory Thomas]

We have a VA7400 with 3TB divided into 2 redundancy groups. According to "armdsp", each group has 133 Gb - Used as Active Hot Spare and 246Gb - Used for Redundancy. I need that space for real data! How can I manipulate those thresholds (or whatever they're called)? Thanks.....Jim

[Enes Dizdarevic]

You can not change space reserved for redudancy. It is how VA works. It has calculated space nedeed for redudancy based on VA mode: raid1 or autoraid.
You can change space reserved for Active Hot Spare. If you have not changed Active Hot Spare mode yet, it is in default mode called Automatic which means: VA will reserve space of largest disk for hot spare per redudancy group if you have less than 15 disks per RG, or space of 2 largest disk per RG if you have more than 15 disks per RG. This can be changed to 0 or 1 largest disk.
CommandView command is armmgr option -h
None or
Automatic or
LargestDisk or
LargestTwoDisk.

[Gregory Thomas]

Thanks Enes....armmgr -h none got rid of the hot spare freeing up about 260 Gb. Redundancy is holding almost 500Gb. That's 1/6 of the whole array! This must be something a salesman invented.

[Vincent Fleming]

Scott,

The redundancy is the RAID overhead. For RAID 1 and 0/1, it's 50% (the data is mirrored). For RAID5DP, it's dependent on the number of drives (and their size) in the redundancy group - it always uses 2 (effectively).

How it (RAID5DP) really works is for each stripe of data on the RG, it keeps 2 parity blocks (one XOR parity and the other CRC). RAID5Dp generally uses significantly less space for redundancy than RAID0/1.

If you're not using AutoRAID, you can recover a lot of that redundancy space by switching to AutoRAID.

The net of all this is: you WANT that space for redundancy - otherwise, a disk going bad will destroy your data!

BTW: I don't suggest you run without active spares. If you have a disk fail, the unit will rebuild your data on the spare. With no spare, and enough disk failures, you *will* lose all your data.

Good luck!

[Ralf Bressel]

if you want no redundancy, you shold use has-modules and raid 1. expect redundancy of about 0.01 % ...

[Ralf Bressel]

sorry, instead of raid 1 read:
raid 0