Solaris

Hi,

We had a disk go bad in this SunFire 280R, running Solaris 8 - Generic_117350-46. We replaced the disk and partitioned it. We metadetached all the metadevices but for some reason the new disk is showing up in the metadb as Concat/Stripe and it won't let us re-sync the mirror. We are not Sun DiskSuite experts here and need some help. Here is the full metastat:

sl76prd2# metastat d35
d35: Mirror
Submirror 0: d15
State: Okay
Pass: 1
Read option: roundrobin (default)
Write option: parallel (default)
Size: 8389656 blocks

d15: Submirror of d35
State: Okay
Size: 8389656 blocks
Stripe 0:
Device Start Block Dbase State Hot Spare
c1t0d0s5 0 No Okay


sl76prd2# metastat
d30: Mirror
Submirror 0: d10
State: Okay
Pass: 1
Read option: roundrobin (default)
Write option: parallel (default)
Size: 2941002 blocks

d10: Submirror of d30
State: Okay
Size: 2941002 blocks
Stripe 0:
Device Start Block Dbase State Hot Spare
c1t0d0s0 0 No Okay


d31: Mirror
Submirror 0: d11
State: Okay
Submirror 1: d21
State: Needs maintenance
Pass: 1
Read option: roundrobin (default)
Write option: parallel (default)
Size: 4194828 blocks

d11: Submirror of d31
State: Okay
Size: 4194828 blocks
Stripe 0:
Device Start Block Dbase State Hot Spare
c1t0d0s1 0 No Okay


d21: Submirror of d31
State: Needs maintenance
Invoke: metareplace d31 c1t1d0s1 <new device>
Size: 4194828 blocks
Stripe 0:
Device Start Block Dbase State Hot Spare
c1t1d0s1 0 No Maintenance


d33: Mirror
Submirror 0: d13
State: Okay
Pass: 1
Read option: roundrobin (default)
Write option: parallel (default)
Size: 8389656 blocks

d13: Submirror of d33
State: Okay
Size: 8389656 blocks
Stripe 0:
Device Start Block Dbase State Hot Spare
c1t0d0s3 0 No Okay


d34: Mirror
Submirror 0: d14
State: Okay
Pass: 1
Read option: roundrobin (default)
Write option: parallel (default)
Size: 8389656 blocks

d14: Submirror of d34
State: Okay
Size: 8389656 blocks
Stripe 0:
Device Start Block Dbase State Hot Spare
c1t0d0s4 0 No Okay


d35: Mirror
Submirror 0: d15
State: Okay
Pass: 1
Read option: roundrobin (default)
Write option: parallel (default)
Size: 8389656 blocks

d15: Submirror of d35
State: Okay
Size: 8389656 blocks
Stripe 0:
Device Start Block Dbase State Hot Spare
c1t0d0s5 0 No Okay


d37: Mirror
Submirror 0: d17
State: Okay
Pass: 1
Read option: roundrobin (default)
Write option: parallel (default)
Size: 16779312 blocks

d17: Submirror of d37
State: Okay
Size: 16779312 blocks
Stripe 0:
Device Start Block Dbase State Hot Spare
c1t0d0s7 0 No Okay


d23: Concat/Stripe
Size: 8389656 blocks
Stripe 0:
Device Start Block Dbase
c1t1d0s3 0 No

d24: Concat/Stripe
Size: 8389656 blocks
Stripe 0:
Device Start Block Dbase
c1t1d0s4 0 No

d25: Concat/Stripe
Size: 8389656 blocks
Stripe 0:
Device Start Block Dbase
c1t1d0s5 0 No

d27: Concat/Stripe
Size: 16779312 blocks
Stripe 0:
Device Start Block Dbase
c1t1d0s7 0 No

d38: Concat/Stripe
Size: 185006080 blocks
Stripe 0:
Device Start Block Dbase
c3t0d0s0 0 No
Stripe 1:
Device Start Block Dbase
c3t0d0s3 0 No

d39: Concat/Stripe
Size: 7340032 blocks
Stripe 0:
Device Start Block Dbase
c3t0d0s4 0 No

d41: Concat/Stripe
Size: 4200448 blocks
Stripe 0:
Device Start Block Dbase
c3t0d0s1 0 No

hsp001: is empty

Any help appreciated!

Thanks,

Mike

This should be pretty simple to sort out but I'm going to have to make a few assumptions as I don't know your metadevice naming standard, you've not posted your md.tab and without CODE tags round the metastat output it's very hard to read.
</grumble>

Righty, that out of the way, lets take a look at your current status and draw some inferences:
I think your current setup is thus:
Mirror devices:
d37
d35
d34
d33
d31
d30
Each mirror has two submirrors, d1x and d2x. Does that sound correct?
Your metadb is on slice 6, correct?
It also looks like the disk that died is c1t1d0, correct?

If all those assumptions are correct, (if not, stop here and clarify) then the normal course of action is to replace the failed disk, then run metareplace -e <mirror device> c1t1d0s<slice number> for each slice of the new disk.
Unfortunately you've already blown away most of your submirrors so we'll need to rebuild - don't worry, that's safe and easy and you won't lose any data Just a little more tedious.
As you've left d31 intact, fix that one by running:

Code:

metareplace -e d31 c1t1d0s1

While that rebuilds, we'll check the metadbs out.
Run

Code:

metadb

to list all the copies of the db. Anything with a W by it (probably everything on c1t1d0s6) is buggered. Provided you still have a few good copies of it, remove the faulty versions of the db. Count how many there are on that device, then run:

Code:

metadb -d c1t1d0s6

.
Then re-add them:

Code:

metadb -a -c <number of copies> c1t1d0s6

.
Now wait here until d31 has finished synching:

Code:

metastat | grep '%'

Wait until it stops showing you a % complete.
As you've already created all your submirror metadevices, the only thing left to do it attach them - One by one (yes really, anything else is _way_ slower). For each remaining mirror in the list, run:

Code:

metattach d3x d2x

I don't know what d38 d39 and d41 are for so I'm ignoring them for now

Thanks for the reply. Sorry for not including "CODE tags". The first assumption is correct - each mirror has two submirrors, d1x and d2x.

The metadb is on s6. The disk c1t1d0 is the replaced disk.

I was trying to execute:

Code:

metareplace -e d30 c1t1d0s0

and it kept barfing. When I executed:

Code:

metareplace -e d31 c1t1d0s1

It started syncing! Thanks. But the metadb is messed up like you said:

Code:

# metadb
        flags           first blk       block count
      W   p  l          16              1034            /dev/dsk/c1t1d0s6
      W   p  l          1050          1034            /dev/dsk/c1t1d0s6

I tried the command:

Code:

# metadb -d c1t1d0s6
metadb: sl76prd2: must have at least 2 databases (-f overrides)

Suggestions?

Thanks,

Mike