lvmattachd process

[Christine Hartman]

I have a process called lvmattachd running on the system. This process is not a familiar one to me. Has anyone ever heard of it, and if so can you tell me what it is and what does it do?

[Edgar_8]

Hi,

Could you provide the output of the following:

# ps -ef | grep lvm

Regards

[Steven E. Protter]

LVM processes have to do with logical volume management. Disk management.

If there are lvm processes with little cpu time all is well. If your system has high i/o come lvm processes will show high cpu time.

It is a possible sign of disk trouble when lvm processes are taking a lot of resources. It could mean there is a disk problem, or it could mean you need lvm patches.

There are a lot of lvm patches produced and I recommend you keep current. I've made several strange situations go away merely by putting in a newer set of lvm patches.

SEP

[Christine Hartman]

here is the ps -ef output. Because of the low PID I'm assuming these processes are supposed to be running..and no they are not taking up much CPU at all. Some DBA asked and wanted to make sure it's not interfering with his superior database processes :P (can you say paranoid?). I'm wondering if there is a correlation between how many of these run to the number of VGs I have created on the system?

>ps -ef |grep lvm
root 19 0 0 Nov 15 ? 0:09 lvmkd
root 20 0 0 Nov 15 ? 0:12 lvmkd
root 21 0 0 Nov 15 ? 0:09 lvmkd
root 22 0 0 Nov 15 ? 0:10 lvmkd
root 23 0 0 Nov 15 ? 0:09 lvmkd
root 24 0 0 Nov 15 ? 0:09 lvmkd
root 25 0 0 Nov 15 ? 0:00 lvmschedd
root 36 0 0 Nov 15 ? 0:08 lvmdevd
root 37 0 0 Nov 15 ? 53:30 lvmattachd
root 427 0 0 Nov 15 ? 0:00 lvmattachd
root 420 0 0 Nov 15 ? 0:00 lvmdevd
root 422 0 0 Nov 15 ? 0:00 lvmdevd
root 421 0 0 Nov 15 ? 0:00 lvmattachd
root 425 0 0 Nov 15 ? 0:00 lvmattachd
root 423 0 0 Nov 15 ? 0:00 lvmattachd
root 424 0 0 Nov 15 ? 0:00 lvmdevd
root 426 0 0 Nov 15 ? 0:00 lvmdevd
root 428 0 0 Nov 15 ? 0:00 lvmdevd
root 429 0 0 Nov 15 ? 0:00 lvmattachd

[Devender Khatana · Accepted solution]

Hi,

All these are normal processes.

Except lvmattachd everything runs on every system.

Perhaps lvmattachd runs on system with pvlinks. I do not have any such system to check with. Does your system has pvlinks. Confirm by

#vgdisplay -v vgname |grep -i link

HTH,
Devender

[Christine Hartman]

yes...we do have pvlinks....that makes sense to me....

[Christine Hartman]

although....some systems that also have PVlinks do not have this lvmattachd process running. Hmmm.

[Devender Khatana]

Hi,

Are some alternate links active on the system which have these processes running? (Means primary link has failed)

HTH,
Devender

[A. Clay Stephenson]

Here is my best guess although you would really have to consult someone with knowledge of the LVM internals for a definitive answer.

Look at the release notes for PHKL_33390 (although the daemon was actually added at an earlier patch level).

---------------------
( SR:8606324162 CR:JAGae86627 )
When a PV is not accessible at the time of activation, it
will be marked NOT_ATTACHED and no user I/O will be done on
that PV. The PV will unnecessarily remain in that state
till the next (re)activation even if it becomes available.

Resolution:
Introduced a new kernel daemon which will keep track of
NOT_ATTACHED PVs and attach them to the VG as soon as they
become accessible. This will avoid the need to reactivate.
-----------------------------------

[Christine Hartman]

thanks all I think I have sufficient information to reply back to this guy.

[James R. Ferguson]

Hi Christine:

The patch provided by Clay is actually a very useful patch to add to an 11.11 system! It's part of the implementation of OLR (Online LVM Disk Replacement).

See the Technical Knowledge Base document #4000036382 for more information.

PKHL_33390 is the succsssor patch to PHKL_31216 and PHCO_33310 is the successor to PHCO_30698 mentioned therein.

Regards!

...JRF...

[Torsten.]

The patches mentioned above are bringing the ability to "switch on/off" the disks to enhance the ability for online replacement.

for more information:

http://docs.hp.com/en/5991-1236/When_Good_Disks_Go_Bad.pdf