Tek-Tips is the largest IT community on the Internet today!

Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!

  • Congratulations SkipVought on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!

Disksuite & Solaris 7

Status
Not open for further replies.

Igaduma

Technical User
Nov 23, 2001
322
0
0
BE
Hi all,

I need to hotswap a faulty hd on a 220R running Solaris7.
it's a 2way mirror setup on c0t0d0/c0t1d0
Some 2nd submirros are broken and others still work, but it seems time to replace the disk.
I'm used to working with 8/9/10 > Solaris7 for these types of operations, but never really hot-swap.
Generally I asses the system, take some backups, remove the metadb's from the faulty hd, break the 2nd submirrors, stop the system, insert a new disk, dicover it, toc the table, create metadb's & setup the 2nd submirrors again, attach & resync.
But thats on anything > solaris 7 and not hotswapping.

I can't find the hotswap instructions for Solaris7 anymore on docs.sun but I presume I'll be using some cfgadm commands.

If anyone has any advice/warnings/good url, please do share the happiness :)

Thanks!
Iga
 
what do you mean by stop the system? you do not shutdown the servers?

my only suggestion is when discovering the drive, since you are familiar with Solaris 8/9/10, you are spoiled with the command: devfsadm.

I think you will have to use: drvconfig to discover the drives (there is no devfsadm in Solaris 7). like you it has been a few years since I have used Solaris 7, so trying to remember the gotchas is hard.

everything else seems like you have it under control, although I am still wondering what you meant by 'stop the system'


David

 
Hi David,

I could use cfgadm to bring the disk offline for replacing while the system is still running, using the other disk, but I'm not sure if it works if there is just 1 scsi controller that has the c0t0d0s0 root disk/slice attached.
I would therefore feel more comfortable halting the system, powering it down ,replace the disk and perform a boot -r
But again, I'm not sure how many of the normal arsenal of commands is available on Solaris7.
And, before the intervention there is ofcourse the good old ufsdump to tape of each slice before the action begins...for those "ouch..." moments :)

Another question:
I'll be packing a DDS3 sun tape device, I presume all the commands such as "tapes, mt" etc are available on Solaris7 ?

Thanks!
 
As far as I know they should be, as they predate Sol. 7 by some years!

I want to be good, is that not enough?
 
a very good point which escaped me :)
dd, cpio, tar, ... all probably even older as me!
 
Hmm,

With cfgadm, if the system does not see the disk in the first place than that will not work.if I remember right drvconfig and a few other commands were all rolled up in to one wonderful command: devfsadm. Used to rediscover new hardware or changes in hardware. Basically the same as boot -r (well almost).

Providing you take your time and think about it, the risk is not high. You are only detaching a disk that probably is not functioning already. You still have one good side, and as long as you do not make any typos,and mess with your good disk, you will be fine. Word of caution if you chose to power down the server, make sure you delete the metadb replicas for the bad disk before powering it down, and/or add the line in your /etc/system that let's you run on less than 50% + 1 replicas. If you do not do one of these, you probably will go right into maintenance mode upon bootup since you do not have enough functioning replicas.

Don't get me wrong, if you can be very careful and take backups and or other steps to take caution, I am always in favor of taking the safe route!!

 
Hi djr111,

Thanks for the hint, it is indeed better to do that in /etc/system.
I would also add set md_mirror:md_resync_bufsz = 2048 to let it sync faster, but I'm not sure if a solaris7/220R can cope with that.


 
Hi all,

all well, but 5.7 broke on set md_mirrored_root , didn't reckognize it, so I ended up going into maintenance mode, destroying the faulty disk submirrors, and recreating them on the new disk.
Note to everyone: take a ufsdump backup... /var was corrupt after the ufsdump for some reason & fsck -F ufs -o b=<number> didn't work...ufsrestore /var slice and sito-presto, a happpy server!

Thanks for the advice,
L


 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Back
Top