Hi,

I have 4 HD's (hot swapables) running a RAID 5, so 2 HD mirror the other 2
HD's. Today Drive number 2 probably died because my TC4100 server is beeping
and the drive itself is showing a Orange led instead of the green flashing
ones. I have a spare machine with the same specs just sitting in case of an
emergency, What would happend if I swap the dead drive with the spare one if
I'm running a SCO 5.0.5 computer without shutting down the computer ? Would
it kill the file system ? Is it better to wait until the end of the day and
do the operation or just do it now ?

Thanks

Corrlens wrote:
> Hi,
>
> I have 4 HD's (hot swapables) running a RAID 5, so 2 HD mirror the
> other 2 HD's. Today Drive number 2 probably died because my TC4100
> server is beeping and the drive itself is showing a Orange led
> instead of the green flashing ones. I have a spare machine with the
> same specs just sitting in case of an emergency, What would happend
> if I swap the dead drive with the spare one if I'm running a SCO
> 5.0.5 computer without shutting down the computer ? Would it kill the
> file system ? Is it better to wait until the end of the day and do
> the operation or just do it now ?
>
> Thanks

Assuming they really are hot-swap bays and not just removable bays (If they
are SCA or SCA2 (80 pin) connector drives, that removes all doubt, but there
are some actual hot-swappable bays that take ordinary scsi drives and look
the same as non-hot-swappable but merely front-panel removeable bays.)
Just yank the bad drive out and pop in one from the standby machine and
order a replacement.
No shutdown or anything. Once the drive is marked bad it's already offline
and pulling it out of the machine makes no difference.

If it's not really hot-swap though, then you need to shut down because
disconnecting or connecting anything to/from the bus while it's up & running
may crash the scsi bus and possibly corrupt the other disks.

If the bays are as large as a 5 1/2" (half-height) drive, then there is a
small chance they are merely easily removeable and not necessarily
hot-swappable. If the bays are only a little larger than the 3 1/2" drive
itself, then they are almost certainly sca2 80-pin hot-swap drives.

One thing that can't be answered yet is if the new drive will automatically
start rebuilding or if you will have to run a utility to initiate the
rebuild. It's possible that there is no daemon monitoring the raid array to
detect new drives and initiate rebuilds, and it's possible there is no
run-time utility installed to do it manually either, in which case you'd
need to shutdown and enter the raid card bios during boot up. However it's
unlikely. generally, for most cards with their drivers installed normally,
new drives are detected and start rebuilding automatically. It might take a
few minutes before the daemon notices and starts the rebuild, since it might
only check the status at 30 second or 1 minute or longer intervals rather
than continuously, so when you plug the new drive in, give it 5 minutes
before you assume it's not working. It might also see the drive and start
rebuilding without showing or telling you in any obvious way. Disk status
lights, if available, are a pretty good clue. It also might not stop beeping
until the rebuild is completed and that might take a couple hours. If the
server slows down after putting the new drive in, that's another good clue
that a rebuild is in progress.

What model is the raid card? Then we can tell you what program to look for
to manually check the status of the array and/or manually initiate a
rebuild, and tell you what to expect in the way of automatic operation and
console messages and syslog entries, etc...

--
Brian K. White -- brian@aljex.com -- http://www.aljex.com/bkw/
+++++[>+++[>+++++>+++++++<<-]<-]>>+.>.+++++.+++++++.-.[>+<---]>++.
filePro BBx Linux SCO Prosper/FACTS AutoCAD #callahans Satriani

"Brian K. White" <brian@aljex.com> wrote in message
news:P8KdnRm3beKmgY3cRVn-uw@comcast.com...
> Corrlens wrote:
> > Hi,
> >
> > I have 4 HD's (hot swapables) running a RAID 5, so 2 HD mirror the
> > other 2 HD's. Today Drive number 2 probably died because my TC4100
> > server is beeping and the drive itself is showing a Orange led
> > instead of the green flashing ones. I have a spare machine with the
> > same specs just sitting in case of an emergency, What would happend
> > if I swap the dead drive with the spare one if I'm running a SCO
> > 5.0.5 computer without shutting down the computer ? Would it kill the
> > file system ? Is it better to wait until the end of the day and do
> > the operation or just do it now ?
> >
> > Thanks
>
> Assuming they really are hot-swap bays and not just removable bays (If
they
> are SCA or SCA2 (80 pin) connector drives, that removes all doubt, but
there
> are some actual hot-swappable bays that take ordinary scsi drives and look
> the same as non-hot-swappable but merely front-panel removeable bays.)
> Just yank the bad drive out and pop in one from the standby machine and
> order a replacement.
> No shutdown or anything. Once the drive is marked bad it's already offline
> and pulling it out of the machine makes no difference.
>
> If it's not really hot-swap though, then you need to shut down because
> disconnecting or connecting anything to/from the bus while it's up &
running
> may crash the scsi bus and possibly corrupt the other disks.
>
> If the bays are as large as a 5 1/2" (half-height) drive, then there is a
> small chance they are merely easily removeable and not necessarily
> hot-swappable. If the bays are only a little larger than the 3 1/2" drive
> itself, then they are almost certainly sca2 80-pin hot-swap drives.
>
> One thing that can't be answered yet is if the new drive will
automatically
> start rebuilding or if you will have to run a utility to initiate the
> rebuild. It's possible that there is no daemon monitoring the raid array
to
> detect new drives and initiate rebuilds, and it's possible there is no
> run-time utility installed to do it manually either, in which case you'd
> need to shutdown and enter the raid card bios during boot up. However it's
> unlikely. generally, for most cards with their drivers installed normally,
> new drives are detected and start rebuilding automatically. It might take
a
> few minutes before the daemon notices and starts the rebuild, since it
might
> only check the status at 30 second or 1 minute or longer intervals rather
> than continuously, so when you plug the new drive in, give it 5 minutes
> before you assume it's not working. It might also see the drive and start
> rebuilding without showing or telling you in any obvious way. Disk status
> lights, if available, are a pretty good clue. It also might not stop
beeping
> until the rebuild is completed and that might take a couple hours. If the
> server slows down after putting the new drive in, that's another good clue
> that a rebuild is in progress.
>
> What model is the raid card? Then we can tell you what program to look for
> to manually check the status of the array and/or manually initiate a
> rebuild, and tell you what to expect in the way of automatic operation and
> console messages and syslog entries, etc...
>

Thanks for answering, I have the HP 1M Raid controller.

corrlens wrote:
> "Brian K. White" <brian@aljex.com> wrote in message
> news:P8KdnRm3beKmgY3cRVn-uw@comcast.com...
>> Corrlens wrote:
>>> Hi,
>>>
>>> I have 4 HD's (hot swapables) running a RAID 5, so 2 HD mirror the

>>
>> What model is the raid card? Then we can tell you what program to
>> look for to manually check the status of the array and/or manually
>> initiate a rebuild, and tell you what to expect in the way of
>> automatic operation and console messages and syslog entries, etc...

>
> Thanks for answering, I have the HP 1M Raid controller.

As it happens, I'm not familliar with that one's particulars at all.
Good luck

--
Brian K. White -- brian@aljex.com -- http://www.aljex.com/bkw/
+++++[>+++[>+++++>+++++++<<-]<-]>>+.>.+++++.+++++++.-.[>+<---]>++.
filePro BBx Linux SCO Prosper/FACTS AutoCAD #callahans Satriani

"Brian K. White" <brian@aljex.com> wrote in message
news:zYudnZvyQaSl2I3cRVn-pQ@comcast.com...
> corrlens wrote:
> > "Brian K. White" <brian@aljex.com> wrote in message
> > news:P8KdnRm3beKmgY3cRVn-uw@comcast.com...
> >> Corrlens wrote:
> >>> Hi,
> >>>
> >>> I have 4 HD's (hot swapables) running a RAID 5, so 2 HD mirror the
>
> >>
> >> What model is the raid card? Then we can tell you what program to
> >> look for to manually check the status of the array and/or manually
> >> initiate a rebuild, and tell you what to expect in the way of
> >> automatic operation and console messages and syslog entries, etc...
>
> >
> > Thanks for answering, I have the HP 1M Raid controller.
>
> As it happens, I'm not familliar with that one's particulars at all.
> Good luck

HP TC4100's come with a hot-swap backplane, and if using a 1M NetRAID
controller, you should be sweet to swap it out.

However, your original emssage said "2 HD mirror the other 2 HD's", which
sounds more like RAID 1, rather than RAID 5.

Using the '/etc/megamgr' tool, verify what sort of container is set up
using the 'View/Modify' options, and then set about getting a replacement
drive.

There should be no issue pulling the failed drive and putting the
replacement in.

You shouldn't need to manually-swap a drive from one-slot to another if the
RAID container is correctly set up.

As you said RAID 5 originally, I'd assume 3drives + 1 spare (in which case
your system should still be running in an Optimal state, rather than
'DEGRADED'). If this is the case, there isn't quite as much urgency to get
a replacement drive (as it will just go back in as a spare), as the
container still has the ability to have one more failure.

If it was RAID 5 with all 4 drives in the one container (no spare), then
you need to get a replacement drive urgently. It is the same if you are
running RAID 1. The container will not be able to cope with a 2nd failure.

bkx

"Stuart J. Browne" <stuart@promed.com.au> wrote in message
news:41104d05@dnews.tpgi.com.au...
>
> However, your original emssage said "2 HD mirror the other 2 HD's", which
> sounds more like RAID 1, rather than RAID 5.

Yeh, that does sound like a RAID1.

Pardon if this is topic drift, but I keep seeing posts in this newsgroup
(over the years) about people running RAID5 on 4 drives. AFAIK, RAID5 uses
THREE drives and any of the three drives can fail without problems. The
remaining two drives can operate and rebuild data on the failed drive when
it is replaced. Therefore, isn't it a total waste to run 4 drives in RAID5
configuration? Isn't the 4th drive a "spare to the built-in spare"?

"Bob Meyers" <oregonbob2000 at yahoo.com> wrote in message
news:6uWdnbc2GshuiozcRVn-uA@centurytel.net...
>
> "Stuart J. Browne" <stuart@promed.com.au> wrote in message
> news:41104d05@dnews.tpgi.com.au...
> >
> > However, your original emssage said "2 HD mirror the other 2 HD's",
which
> > sounds more like RAID 1, rather than RAID 5.
>
> Yeh, that does sound like a RAID1.
>
> Pardon if this is topic drift, but I keep seeing posts in this newsgroup
> (over the years) about people running RAID5 on 4 drives. AFAIK, RAID5 uses
> THREE drives and any of the three drives can fail without problems. The
> remaining two drives can operate and rebuild data on the failed drive when
> it is replaced. Therefore, isn't it a total waste to run 4 drives in RAID5
> configuration? Isn't the 4th drive a "spare to the built-in spare"?

RAID 5 requires a MINIMUM of three drives. I have seen up to a ten drive
RAID
5 (there are some that are probably larger) with three hot spares standing
by.

Adding more than three HD's to a RAID 5 just adds more storage space.

I currently run a four drive RAID 5 with two hot spares. At the time I
wanted more
storage space for a single drive (Four 18GB drives acting is 54GB, obviously
separated into multiple partitions).

moncho.