trove resize-volume fails with resize2fs error

[Simon Chang]

When resizing the volume of an instance with the following command:
trove resize-volume <instance_id> <size>

The instance remains in RESIZE status and never return to ACTIVE.

The guestagent log contain this error:
trove.guestagent.volume Error resizing file system.
trove.guestagent.volume Traceback (most recent call last):
trove.guestagent.volume File "/home/stack/trove/trove/guestagent/volume.py", line 124, in resize_fs
trove.guestagent.volume run_as_root=True, root_helper="sudo")
trove.guestagent.volume File "/home/stack/trove/trove/openstack/common/processutils.py", line 200, in execute
trove.guestagent.volume cmd=sanitized_cmd)
trove.guestagent.volume ProcessExecutionError: Unexpected error while running command.
trove.guestagent.volume Command: sudo resize2fs /dev/vdb
trove.guestagent.volume Exit code: 1
trove.guestagent.volume Stdout: u''
trove.guestagent.volume Stderr: u"resize2fs 1.42 (29-Nov-2011)\nPlease run 'e2fsck -f /dev/vdb' first.\n\n"

Changing the following line in trove/guestagent/volume.py from ..
utils.execute("e2fsck", "-f", "-n", self.device_path,
              run_as_root=True, root_helper="sudo")

to ..

utils.execute("e2fsck", "-f", "-p", self.device_path,
              run_as_root=True, root_helper="sudo")

.. resolved the issue.

The e2fsck flag was changed from -p to -n in this patch:
https://review.openstack.org/#/c/127629/

We need to revisit the patch.

[Nikhil Manchanda (slicknik)]

Any updates on this?
If not, I will probably end up moving this out to kilo-2.

[Simon Chang (changsimon)]

Nikhil, I don't have a patch for this one yet. Ubuntu has this issue, but Centos don't, and I'm trying to test with Debian now.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/140488

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/140488
Committed: https://git.openstack.org/cgit/openstack/trove/commit/?id=a13112599dd20e890d8d70babb0015052796aa09
Submitter: Jenkins
Branch: master

commit a13112599dd20e890d8d70babb0015052796aa09
Author: Simon Chang <email address hidden>
Date: Tue Dec 9 16:49:03 2014 -0500

    Fix trove resize-volume resize2fs error

    trove resize-volume works by calling the e2fsck and resize2fs
    commands. When resize2fs detect errors in the volume, it will
    abort with an error: "Please run 'e2fsck -f /dev/vdb' first".

    The e2fsck command is currently called with the -n flag, meaning
    not to fix any volume errors, which cause the subsequent resize2fs
    command to fail. As a result, the trove instance being resized
    remain stuck in the RESIZE status.

    This patch reverts the e2fsck -n flag back to the -p flag, which
    enables auto correction of volume errors, thus allowing resize2fs
    to complete the resize-volume operation.

    Change-Id: I14b43966d4e7f047a5859bd4a463f2838f03b01f
    Closes-Bug: #1391639