[Starlingx-discuss] StarlingX 4.0 Issue in Extending XFS Filesystem

Anirudh Gupta anyrude10 at gmail.com
Wed Dec 16 14:54:00 UTC 2020


Hi Team,

We are Installing StarlingX 4.0 and facing an issue which is related to the
below bug:

https://bugs.launchpad.net/starlingx/+bug/1904973

We have Raid 5 configuration at our setup on which STX 4.0 is installed.

While extending the XFS filesystem for docker-lv before uploading
stx-openstack application

https://docs.starlingx.io/deploy_install_guides/r4_release/openstack/install.html

Command:

*system host-fs-modify controller-0 docker=60*

As we run this command, we get kernel alert logs in */var/log/kern.log/*


2020-12-16T13:23:40.467 controller-0 kernel: alert [  921.174512] XFS
(dm-5): metadata I/O error in "xfs_trans_read_buf_map" at daddr 0x77fc002
len 1 error 74
2020-12-16T13:23:41.715 controller-0 kernel: alert [  922.488879] XFS
(dm-5): Metadata CRC error detected at xfs_agi_read_verify+0xd0/0xf0 [xfs],
xfs_agi block 0x77fc002
2020-12-16T13:23:41.715 controller-0 kernel: alert [  922.498188] XFS
(dm-5): Unmount and run xfs_repair
2020-12-16T13:23:41.715 controller-0 kernel: alert [  922.507214] XFS
(dm-5): First 128 bytes of corrupted metadata buffer:
2020-12-16T13:23:41.809 controller-0 kernel: alert [  922.516545] 00000000:
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
2020-12-16T13:23:41.809 controller-0 kernel: alert [  922.516547] 00000010:
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
2020-12-16T13:23:41.809 controller-0 kernel: alert [  922.516548] 00000020:
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
2020-12-16T13:23:41.809 controller-0 kernel: alert [  922.516549] 00000030:
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
2020-12-16T13:23:41.809 controller-0 kernel: alert [  922.516551] 00000040:
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
2020-12-16T13:23:41.809 controller-0 kernel: alert [  922.516552] 00000050:
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
2020-12-16T13:23:41.809 controller-0 kernel: alert [  922.516553] 00000060:
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
2020-12-16T13:23:41.809 controller-0 kernel: alert [  922.516555] 00000070:
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................

In order to avoid this, we thought to initially set the size of docker-lv
filesystem to 60 instead of increasing it from 30 to 60

For this, before bootstrap C0, we changed the following:

1. lv_size changed from 30 to 60 in file
*/usr/share/puppet/modules/platform/manifests/filesystem.pp*

class platform::filesystem::docker::params::bootstrap (
  $lv_size = '*60*',
  $lv_name = 'docker-lv',
  $mountpoint = '/var/lib/docker',
  $devmapper = '/dev/mapper/cgts--vg-docker--lv',
  $fs_type = 'xfs',
  $fs_options = '-n ftype=1',
  $fs_use_all = false
) { }

2. Value of KUBERNETES_DOCKER_STOR_SIZE  at line 39 changed to 60 from 30
in file */usr/lib64/python2.7/site-packages/sysinv/common/constants.py*

KUBERNETES_DOCKER_STOR_SIZE = *60*

With these changes, after bootstrap of Controller-0, the size of docker-lv
changes to 60 successfully.
The same is verified running the command "system host-fs-list controller-0"

But once Controller-1 bootstraps and comes up, we still get the size of
docker-lv as 30.
Since Controller-0 PXE boots Controller-1, there must be a file at C0 which
we could change so that Controller-1's docker-lv gets a size of 60G.

We are unable to find such a file and would require help in order to
resolve this.

Regards
Anirudh Gupta
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.starlingx.io/pipermail/starlingx-discuss/attachments/20201216/a41de958/attachment-0001.html>


More information about the Starlingx-discuss mailing list