0

I have taken over a server with a 3ware 9650SE-8LPML raid controller, the server has 8 disks and I configured it to be a RAID 10, no spares. My questions is is it normal that when you start the verify it is doing first 6 disks and almost at the end of the verify process it is doing the last 2 disks.

It verifies disk 0[phy5], 1[phy6], 2[phy7] with disk 5[phy2], 6[phy3], 7[ph4] and then the disk 3[phy0] with 4[phy1].

I have to admit that I am not sure if this is the correct order for each 4 disks wired up to each port on the controller, but it just seems a bit odd to me how it verifies the raid, can someone please confirm this is normal and ok?

Is it correct to think that for each set of 4 disks that they should be on their own cable to the controller?

Update: the raid set is configured and after the reboot it shows 4 subunits:

8 drive 64K x.xx TB (RAID10)

phy7 with phy6

phy5 with phy4

phy3 with phy2

phy1 with phy0

Any help much appreciated.

Nick
  • 3
  • 2
  • I can't find any documentation stating so, but I seem to recall that the 3ware controllers won't do more than 3 simultaneous verify/rebuilds concurrently. Each one of the subunits is a separate task. –  Aug 15 '14 at 22:22

1 Answers1

1

The verification pattern is a bit odd, but not entirely unexpected depending on the controller's CPU capacity and other limitations. This particular controller can be configured to slow the verification process as not to interfere with normal usage.

In my experience, I have seen controllers checking only 2 disks simultaneously while other controllers with check a 24 array at once (the performance impact was considerable).

Go to Management > Controller Settings in 3BM and play with the Faster I/O vs. Faster Verify rate to see if the pattern changes, but I wouldn't worry unless it's causing problems for you.

gtirloni
  • 5,746
  • 3
  • 25
  • 52
  • 1
    Thanks a lot for your time to explain, I start to like the controller as it seems pretty clever, I did the new config in the controller itself, I had read the manual at least twice but could not find any explanation for this behaviour. Great, I can now continue and try to put VMware on it 'another adventure'. – Nick Aug 16 '14 at 10:29