vgimport error

[William Pribble]

I am running an establish/split on are DMX frame and running into a error. The standards and BCVs are attached to the same server.

The volume groups and file systems are /dev/vg18/lvol1 /oraxx, and /dev/vg18_bcv/lvol1 /oraxx_bcv

Here are the commands that I am using.


umount /oraxx_bcv
vgchange -a n /dev/vg18_bcv
symmir -g symdg_oraxx establish -full -exact -noprompt


After the establish is complete:

umount /oraxx
vgchange -a n /dev/vg18
symmir -g symdg_oraxx split -noprompt

After split completes:

vgexport /dev/vg18_bcv
vgchgid /dev/rdsk/c20t5d3 /dev/rdsk/c20t5d4

vgexport -v -p -m /etc/lvmconf/vg18.mapfile /dev/vg18

vgchange -a y /dev/vg18
mount /oraxx
mkdir /dev/vg18_bcv
mknod /dev/vg18_bcv/group c 64 0x080000
vgimport -m /etc/lvmconf/vg18.mapfile -f /usr/local/admin/vg18_bcv.vgimport /dev/vg18_bcv


This is where it bombs out.
vgimport: Warning: Volume Group contains "60" PVs, "4" specified. Continuing.
vgimport: Quorum not present, or some physical volume(s) are missing.


Just remove the symmir commands and I can run the exact commands in the exact sequence and everything works fine. I am able to vgchange /dev/vg18_bcv and mount /oraxx_bcv


Any help would be greatly appreciated!

[Kent Ostby]

I think you need to create your vgmap prior to the initial split.

Then use that map after you do the resync.

That way you are using a map when the VG contains all its volumes when you try to set it back up that way at the end.

[Ninad_1]

Hi,

How many PVs do you actually have in your VG.
If its more than 2 dont you need to include them all at the same time while doing the vgchgid.

Regards,
Ninad

[William Pribble]

There is only 2 devices. That's why I trying to figure out where the 60 is coming from?

[William Pribble]

Kent,

I tried your suggestion but go the same result.

This sequence has worked for years on our EMC 8830, were moving to the DMX3 and now we are getting this error.

[Ninad_1]

Jeffrey,

I am sure you must have taken care of this -
the group file should have unique minor no.
ls -l /dev/*/group to verify this.

strings /etc/lvmtab to confirm PVs in /dev/vg18

and that the disks you have specified are indeed the BCV disks.

Regards,
Ninad

[William Pribble]

Ninad,

Unfortunatley I have checked all those, and if there were duplicate minor numbers my script would stop.

Thanks,
Jeff

[sathish kannan]

Hello Jefrey,
Your steps looks ok to me. Is this first time you are trying establish and split on these devices?

Have you tried exporting source devices VG (vg18) before changing the "vgchgid"?

Why do you use -f option with vgimport command? Is it to specify alternate links?
If yes, there may chance with the order of devices in Primary and alternate link? Why don't you just try with out -f option?

BTW, I guess you have checked your lvmtab file and symdg and selected correct devices?

Regards
Sathish

[Ninad_1]

Jeffrey,

I dont know whether this will help, but can you try this sequence. I have suggested few changes/steps. Is the file /usr/local/admin/vg18_bcv.vgimport updated and points to same BCV disks ?

1.umount /oraxx_bcv
2.vgchange -a n /dev/vg18_bcv
3.vgexport /dev/vg18_bcv
4.symmir -g symdg_oraxx establish -full -exact -noprompt

After the establish is complete:
5.umount /oraxx
6.vgchange -a n /dev/vg18
7.symmir -g symdg_oraxx split -noprompt

After split completes:

8.vgchgid /dev/rdsk/c20t5d3 /dev/rdsk/c20t5d4
9.vgexport -v -p -m /etc/lvmconf/vg18.mapfile /dev/vg18
10.vgchange -a y /dev/vg18
11.mount /oraxx
12.mkdir /dev/vg18_bcv
13.mknod /dev/vg18_bcv/group c 64 0x080000
14.vgimport -m /etc/lvmconf/vg18.mapfile /dev/vg18_bcv /dev/rdsk/c20t5d3 /dev/rdsk/c20t5d4
15.vgchange -a y /dev/vg18_bcv
16.mount /oraxx_bcv

Please let us know the progress. Good luck.
Regards,
Ninad

[William Pribble]

Thanks for all the suggestions! Unfortunately nothing has seemed to help yet.

[Ninad_1]

Hi,

Can you attach the /usr/local/admin/vg18_bcv.vgimport file.
I have a gut feeling that the 1st disk specified in this file has the information that there are 60 PVs in the volume group and the file actually contains 4 disk file names.

Regards,
Ninad

[rariasn]

Hi Jeffrey,

Verify infile: /usr/local/admin/vg18_bcv.vgimport

or

vgimport -m /etc/lvmconf/vg18.mapfile /dev/vg18_bcv /dev/rdsk/c20t5d3 /dev/rdsk/c20t5d4

ran

[William Pribble]

Ninad,

After reading some more information on the forums, I decided to rebuild the standard mount point.

Now I get a different error:

Physical volume "/dev/rdsk/c20t5d3" is not a block special file.

The good thing is that I get the error when I try the import with the est/split and when I don't run it.

I checked ioscan, ran insf -e, pvdisplay and all of the information looks good.

[Ninad_1]

Jefferey,

The error is correct - rdsk is a character device and not a block device.
You should use /dev/dsk/cxtydz and not rdsk

Regards,
ninad

[Ninad_1]

Jefferey,

I assumed that the error you are getting is during vgimport - so you need to use block device naming convention - /dev/dsk/c20t5d3

I hope you have already done a
pvcreate /dev/rdsk/c20t5d3.

Please let know if I have misunderstood your problem.

regards,
Ninad

[William Pribble]

Yeah, it must be to early in the morning. I was pointing to the wrong file. Anyway, I still stuck with the original error.

[William Pribble]

Ninad,

Here is the contents of my vgimport file:

/dev/dsk/c20t5d3
/dev/dsk/c23t5d3
/dev/dsk/c20t5d4
/dev/dsk/c23t5d4

Thanks,
Jeff

[Ninad_1]

Jefferey,

Never mind as long as you are awake now :)
I guess that you are having 2 disks in the VG - and there are 2 alternate links to the disks hence you are mentioning 4 disks in the vgimport file.
Now my first question is are you very much sure that the disk devices you have mentioned are infact the BC copy disks still ? If yes then try manually adding as
vgimport -p /dev/vg18_bcv /dev/dsk/c20t5d3 /dev/dsk/c23t5d3 /dev/dsk/c20t5d4 /dev/dsk/c23t5d4

I am deliberately mentioning the -p flag so that it doesnt actually import but checks all steps it will perform during actual import. See if it shows any errors. Paste the errors here again. If it shows errors it means that some/all of the disks you have mentioned are not actually the BC copy disks.
If it succeds then go ahead and do the vgimport with the mapfile and infile. Assuning you have done the vgchgid etc.

Best of luck,
Ninad

[William Pribble]

I double checked the devices:

/dev/rdsk/c20t5d3 BCV EMC SYMMETRIX 5671 6300291000 8923200
/dev/rdsk/c23t5d3 BCV EMC SYMMETRIX 5671 6300291000 8923200
holldb16|root:/ # syminq |grep -i 292
/dev/rdsk/c20t5d4 BCV EMC SYMMETRIX 5671 6300292000 8923200
/dev/rdsk/c23t5d4 BCV EMC SYMMETRIX 5671 6300292000 8923200

I tried running it the way you suggested and got the same error message.

vgimport: Warning: Volume Group contains "60" PVs, "4" specified. Continuing.
vgimport: Quorum not present, or some physical volume(s) are missing.

Thanks,
Jeff

[Ninad_1]

To be honest I dont know about EMC arrays - so didnt understand what you have shown by syminq - Please can you tell me the output -
what do the c20t5d4 and c23t5d4 indicate - are they different paths to same disk of one is the PVOL/original disk and other is SVOL/BC copy ?
What is happening is that when vgimport accesses the 1st disk in the list it finds that the VG has 60 PVs in it and checks to find that you have mentioned only 4 PVs.
I dont know whether this will be of any use
ioscan -fnC disk
insf -e -C disk

Regards,
Ninad

[sathish kannan]

Hi Jeffrey,
couple of thoughts
1.why don't you just try your script with single path rather than both paths.

2.you mentioned earlier that you have migrated from one EMC frame to another. have you upgraded your solution enabler and inq utility. Just to let you know that output of "inq" is changed. The number of fields for the device from 3 to 5. Please check your scripts whether this change is updated from old output to new.


Ex: for device 6300292000

63 00292 000
serial no. symm symdev Common serialno

regards
Sathish

[usergroup.os.ux-3.4-iro]

Hi,

could it be that your EMC Devicegroup is not properly configured?
it seems for me that your BCV Disks are not the the mirrors of your standard devices.

Check the Devicegroup with symdg show
Perhaps in your Devicegroup there are more than your two disks of vg18 and you are trying to import the wrong disks?

hope it helps,

Peter

[William Pribble]

Thanks for all the responses. There was a problem on the EMC side with one of the devices. I replaced the devices and my scripts ran fine.

[William Pribble]

Problem with EMC device.