Incorrect lvmtab/drives removed improperly

[Steven Buschman_3]

Core issue:Can't see new drive on system being added.

Red flags:I/O errors when looking at disks, filesystems, etc
in SAM, very slow to boot in beginning, 'ioscan' takes 1.5hours to return prompt, but does see the A4999A scsi card
and the new drive on the end (as does searching for boot drives)

The system....
B2000 w/ultrascsi A4999A card running 11.0

Unknown reason why (inherited system) but.....
-strings on lvmtab shows
---vg00 is the boot drive (only one currently working)
---vg01 is c3t1 (not present/attached anymore)
---vg02 is c3d2 (not present/attached anymore)
---vg03 is c3d3 (not present/attached anymore)
---vg04 is c3t4 and dead, clicking, unusable)

-vgdisplay says....
---can't open control file /dev/vg01/group
------(same for vg02 and vg03)
---Cannot display /dev/vg04 because it's not activated

vg04 has no data on it that I care about.
There is no record of anything for vg01,2,3
and as far as I'm concerned, are best gone.
vg04 could go too.
I do want to save the boot drive though, and vg00.

I'm hoping I could manually remove some files and/or issue commands and re-create the lvmtab
so that I can not only get the OS cleaned up,
but see if that is what is making the system so slow and the source of the I/O errors, not
to mention getting a new drive attached.

Any ideas????
Thanks in advance, Steven

[Luc Bussieres_1 · Accepted solution]

Steven,

I would suggest you to first start by recreating your lvmtab file according to what is really attach to the system.

To do so you will first move the file to lvmtab.bak to be able to move it back in place if the new file that you will create is not working properly:

#mv /etc/lvmtab /etc/lvmtab.bak

Then you could scan the system to recreate the file:

#vgscan -v

After that run a vgdisplay to verify if everything is fine and if there is still some error message.

Luc

Then you could remo

[James R. Ferguson]

Hi Steven:

Wait! As you said, "Red flags:I/O errors when looking at disks, filesystems, etc in SAM, very slow to boot in beginning, 'ioscan' takes 1.5hours to return..."

This points to a hardware problem that you need to resolve (bad or loose cable, bad card, duplicate SCSI address, etc.). You do *not* (nor cannot, really) begin to dissect the software level. Doing so, will only make this problem much worse.

Start by taking *off* the new disk and return your configurationt to its original state.

Regards!

...JRF...

[MANOJ SRIVASTAVA]

I Think the new drive added has created the problem , do like thsi remove the same and reboot the sytem you should be able tog et all of the data back . The carefully check for the id's for the new drive and also the cabla and connectors .Please ensure teh termiantions is fine too.


Manoj Srivastava

[Luc Bussieres_1]

Steven,

The steps I gave you previously, where to fix the problem with vg01, vg02 & vg03 that have been remove manually (probably an rm in the directory) with part of them still configured in the /etc/lvmtab file.

It will also be better to remove the Faulty disk first so the vgscan would have more chances to complete successfully. And it may also fix the problem with vg04 if it was created by a conflict in the scsi address.

Luc

[Steven Buschman_3]

Sorry all, I should have posted MORE hardware troubleshooting info, I apologize.

I know the drive currently attached is good (works on another HP and on a SUN).

I have tried 3 different external drives that work on other systems and it has made no difference. I have swapped scsi cables. I have pulled the card out and the system was still slow and taking over an hour to do an ioscan (sam or cmd line). The only other "devices" are boot disk, cdrom, and a fibre card to a brocade with no storage allocated there. I put the scsi card back in and after doing everything a HP online help person recommended hardware-wise (most/all prior to contacting), I have now convinced myself that it is software related. I may be wrong but my tune changed from hardware to software and either way I wanted to clean the system up and get rid of references to non-existing devices.

*** current status***
I moved lvmtab and did the first recommdation
of vgscan -v and it took a while to reboot but
maybe the firbe card does that, and a vgdisplay is clean so far. I am working on the rest from here. Thanks for all the help so far and I'll be back in touch soon.

[Luc Bussieres_1]

Steven,

If you have some I/O errors they are not software related but hardware related.

To know exactly which device is in trouble look at the I/O error message in /var/adm/syslog/syslog.log and should find somthing that look like bdev=0x01100000 or dev_t=0x02010000 I don't recall exactly what is on the left side of the =. But once you have found that error message you could extract the scsi address from the number on the right side:
the 2 first digits are what goes after the c
the next one after the t
and the next one after the d

for example:

for 0x01100000 it will be c1t1d0
for 0x02010000 it will be c2t0d1


Luc

[Steven Buschman_3]

I should be all set, things aren't overall right yet, but the lvm portion looks great.
Still taking almost 2 hours for a ioscan but syslog failures/timeouts poing to yet another missing device (10/1/4/0.xxxx) with 'tldd' erroring so I'm gonna guess that besides the bad disk, a tape library was also removed which is causing the slow ioscans.

Thanks to all for your help, as this is now beyond LVM and if I can't figure it out I'll open one under a different topic header.
Points have been awarded, and a special thanks to Luc who was lightning fast and dead-on.
This forum is far better than other platforms because of everyone on it!