Stubborn Volume Groups

[Don Spare]

I have a problem with volume groups. We recently added some additional EMC disks (including matching BCVs) to our system. I added the masters to the appropriate volume group and extended the filesystem without a problem. Then I applied the same procedure to the volume groups containing the BCVs. But I just noticed that the BCV volumes have not been getting mounted for backup to tape. The error indicated that there was a disk count mismatch. I investigated and found my vgchgid and vgimport commands needed to have the new disks added to the lists of disks to be operated on. I did that but now its all messed up still. I tried, via SAM, to drop the volume groups for the BCVs but SAM complains that there are multiple paths on some devices. Upon further investigation, it appears that disks from volume group vg_bcvDW1 are now part of vg_bcvDW. They never were before because of different disk sizes. Additionally, SAM will not let me remove either volume group. Any ideas will be greatly appreciated. I contemplated editing the vg.conf files til I realized they are not editable. One time I had to do something with LVMTAB but since I am more Oracle DBA than Unix admin I don't play with that too much.

Don Spare

[Paula J Frazer-Campbell]

Hi

mv /etc/lvmtab /etc/lvmtab.bak

vgscan -a

will recreate the lvmrab.



Paula

[Don Spare]

If I do that what is the repercussion to my other volume groups/logical volumes. This is a production system and I don't want anything to go down during the business day.

[Paula J Frazer-Campbell]

Don

This link should give you more info.

http://forums.itrc.hp.com/cm/QuestionAnswer/1,,0x7fc7c8ecad09d6118ff40090279cd0f9,00.html


The move and recreate can be done on a live system, but ensure that you backup first.

Paula

[Don Spare]

After reading the link, I have more questions. We are using dual paths on all of our EMC disks. Device files for primary path are in the c5txdx range and in the c7txdx range for the same disk on the other path. We are using EMC's Powerpath software to manage this. How does this change what I need to do?

Don

[MANOJ SRIVASTAVA]

Hi Don


There are essentially few things which cause this to happens. Multiple paths like if you sre usinfg two paths for the same disk and by mistake you are using coz of typos a alternate path in a differnt volume group .

Also since the errors that you are getting is on the BCV's only it is good since you can refresh it and start from scratch . We had similar problems but it all fizzled to some wrong enteries in the volume groups of the BCV's .The best to deal with it is to check the device files for BCV's correspond to the Physical disks are correct , check whether vgchid and vgimport does run on the volume groups of BCV's , you can may be user -x in the shell script and take the o/p in a log file to exactly know where it is failing .

These can be done online except may be remove the databse going to abckup mode before split or any operations to do with the online env.Incase you want I can send some scripts which we use for the same purpoae.


Manoj Srivastava

[MANOJ SRIVASTAVA]

Hi Don


The powerpath softare doenst interrupt with any of the issues on the disks , it will sense the paths and work work to ablance them , essentially what you need is the corect maaping for the Physical volumes and the BCV's . You can also use some thisrd part to mapp the EMC disks . To know what is the config , you may like to look a the attachment .



Manoj Srivastava