Bug #1636

TM Clone fails with VMFS datastore using SSH

Added by Tino Vázquez almost 8 years ago. Updated over 7 years ago.

Status:ClosedStart date:11/05/2012
Priority:NormalDue date:
Assignee:Tino Vázquez% Done:

0%

Category:Drivers - Auth
Target version:Release 4.0
Resolution: Pull request:
Affected Versions:OpenNebula 3.8

Associated revisions

Revision 73a2e4ad
Added by Tino Vázquez almost 8 years ago

bug #1636: Fix for VMFS TM Clone using SSH

Revision b3bf6236
Added by Javi Fontan almost 8 years ago

Revert "bug #1636: Fix for VMFS TM Clone using SSH"

This reverts commit 73a2e4aded827f90f577ba4947a36c0447754f8e.

History

#1 Updated by Tino Vázquez almost 8 years ago

  • Status changed from New to Closed
  • Resolution set to invalid

This bug is invalid, it comes from a false assumption about the DATASTORE_LOCATION translation done by de core (I believed it didn't get translated for a missconfiguration on my testbed)

#2 Updated by Giovanni Toraldo over 7 years ago

Hi Tino,

can you elaborate on your last comment?

I am facing this problem (the clone command executex via ssh on the remote ESXi is referring the source image with my $ONE_LOCATION path instead of /vmfs/volumes), and I set DATASTORE_LOCATION to /vmfs/volumes in oned.conf.

Which configuration is needed to make it work?

Thanks!

vm.log

Tue Dec 18 12:06:46 2012 [TM][I]: Command execution fail: /srv/cloud/var/remotes/tm/vmfs/clone hydra:/srv/cloud/var/datastores/101/e4b3c182f5bd5d65062bb99e0ba0bf31 vmware001:/vmfs/volumes/100/5/disk.0 5 101
Tue Dec 18 12:06:46 2012 [TM][I]: clone: Cloning /srv/cloud/var/datastores/101/e4b3c182f5bd5d65062bb99e0ba0bf31 in vmware001:/vmfs/volumes/100/5/disk.0
Tue Dec 18 12:06:46 2012 [TM][E]: clone: Command "cd /vmfs/volumes/100/5
Tue Dec 18 12:06:46 2012 [TM][I]: 
Tue Dec 18 12:06:46 2012 [TM][I]: export PATH=/usr/sbin:/sbin:$PATH
Tue Dec 18 12:06:46 2012 [TM][I]: 
Tue Dec 18 12:06:46 2012 [TM][I]: if which sudo > /dev/null; then
Tue Dec 18 12:06:46 2012 [TM][I]: SUDO_CMD="sudo " 
Tue Dec 18 12:06:46 2012 [TM][I]: fi
Tue Dec 18 12:06:46 2012 [TM][I]: 
Tue Dec 18 12:06:46 2012 [TM][I]: $SUDO_CMD vmkfstools -U /vmfs/volumes/100/5/disk.0/disk.vmdk
Tue Dec 18 12:06:46 2012 [TM][I]: $SUDO rm /vmfs/volumes/100/5/disk.0/*
Tue Dec 18 12:06:46 2012 [TM][I]: $SUDO_CMD vmkfstools -i /srv/cloud/var/datastores/101/e4b3c182f5bd5d65062bb99e0ba0bf31/disk.vmdk -d thin /vmfs/volumes/100/5/disk.0/disk.vmdk" failed: Failed to delete virtual disk: The system cannot find the file specified (25).
Tue Dec 18 12:06:46 2012 [TM][I]: rm: cannot remove '/vmfs/volumes/100/5/disk.0/*': No such file or directory
Tue Dec 18 12:06:46 2012 [TM][I]: DiskLib_Check() failed for source disk The system cannot find the file specified (25).
Tue Dec 18 12:06:46 2012 [TM][E]: Error cloning hydra:/srv/cloud/var/datastores/101/e4b3c182f5bd5d65062bb99e0ba0bf31 to vmware001:/vmfs/volumes/100/5/disk.0
Tue Dec 18 12:06:46 2012 [TM][I]: ExitCode: 255
Tue Dec 18 12:06:46 2012 [TM][E]: Error executing image transfer script: Error cloning hydra:/srv/cloud/var/datastores/101/e4b3c182f5bd5d65062bb99e0ba0bf31 to vmware001:/vmfs/volumes/100/5/disk.0

#3 Updated by Giovanni Toraldo over 7 years ago

I tried to make it work for my use case (I need to use vmkfstools because standard vmware driver doing cp gives bad performances over NFS datastores).

I cherry-picked two of your commits, and applied the same logic for LN script too in vmfs drivers:

https://github.com/liberologico/one/commits/one-3.8%2Bvmfs

And it works for me.

What I am missing?

Thanks!

#4 Updated by Tino Vázquez over 7 years ago

Hi Giovanni,

IIRC, this comes from creating the datastore with the wrong DATASTORE_LOCATION set. Try recreating the datastore with the correct DATASTORE_LOCATION.

If this doesn't help, please send to the users mailing list the output of:

  • onedatastore show -x 101
  • onedatastore show -x 102
  • the vm.log of the failed VM
  • oned.log

#5 Updated by Ruben S. Montero over 7 years ago

  • Status changed from Closed to Assigned
  • Resolution deleted (invalid)

Just reopening it, to confirm if the problem persists.

#6 Updated by Ruben S. Montero over 7 years ago

  • Status changed from Assigned to Closed

Also available in: Atom PDF