Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[tests] blkid / mdadm on 5.4+ kernels #906

Open
mator opened this issue Dec 5, 2019 · 4 comments
Open

[tests] blkid / mdadm on 5.4+ kernels #906

mator opened this issue Dec 5, 2019 · 4 comments

Comments

@mator
Copy link
Contributor

mator commented Dec 5, 2019 •
edited
Loading

JFYI

mdadm commit mdadm-4.1-54-g329dfc28de makes some tests fail for util-linux on kernel 5.4+, which use --metadata flag on mdadm command line:

tests/ts# ./blkid/md-raid0-whole
        blkid: MD raid0 (whole-disks)         ... FAILED (blkid/md-raid0-whole)

quick grep on util-linux test suite, finds some files:

$ find . -executable -type f | xargs egrep -l "metadata="
./ts/blkid/md-raid1-whole
./ts/blkid/md-raid0-whole
./ts/blkid/md-raid1-part
./ts/fdisk/align-512-4K-md

testing:

# uname -r
5.4.0-11681-g63de37476ebd

# dpkg -l mdadm
||/ Name           Version      Architecture Description
+++-==============-============-============-==================================================
ii  mdadm          4.1-4        sparc64      tool to administer Linux MD arrays (software RAID)

# mdadm --create /dev/md8 --metadata=0.90 --chunk=64 --level=0 --raid-devices=2 /dev/loop0 /dev/loop1
mdadm: 0.90 metadata does not support layouts for RAID0

$ apt changelog mdadm
mdadm (4.1-4) unstable; urgency=medium

  * QA upload.
  * Add support for RAID0 layouts. (Closes: #944676) (LP: #1850540)

 -- dann frazier <yyy>  Mon, 02 Dec 2019 14:35:49 -0700

mdadm (4.1-3) unstable; urgency=medium

  * QA upload.
  * Adding debconf question to disable daily array scan (Closes: #689304):
    - if enabled (default), mdadm checks once a day for degraded arrays
      and missing spares to ensure that such events don't go unnoticed.
    - this is not useful when the system is part of an infrastructure
      with dedicated monitoring systems (e.g. icinga).
    - this debconf question makes it possible to disable these checks.

 -- Daniel Baumann <xxx>  Fri, 09 Aug 2019 17:19:20 +0000
...

array is not created.

testing with older mdadm version:

# dpkg -l mdadm
||/ Name           Version      Architecture Description
+++-==============-============-============-==================================================
ii  mdadm          4.1-3        sparc64      tool to administer Linux MD arrays (software RAID)

# mdadm --create /dev/md8 --metadata=0.90 --chunk=64 --level=0 --raid-devices=2 /dev/loop0 /dev/loop1
mdadm: array /dev/md8 started.

Why do we insist to use older (0.90) metadata format instead of current one (1.2) ?

tested with x86_64 and sparc64 VMs installed with debian sid/unstable.

PS: don't have time currently to write a proper patch, but leave it here, up until i'll find time or someone writes it before me.
@karelzak
Copy link
Collaborator

karelzak commented Dec 5, 2019

The old version of the metadata is stored at the end of the device -- this is very confusing in some use cases, and we need to test it :-)

It sounds like we need to test if mdadm supports it and if not than skip the test.

@mator mator changed the title [tests] mdadm on 5.4+ kernels [tests] blkid / mdadm on 5.4+ kernels Dec 5, 2019
@mator
Copy link
Contributor Author

mator commented Jan 6, 2020

probably test for kernel version (5.4+), and if it raid0 , then not to use old metadata ?! What logic should we use?

karelzak added a commit that referenced this issue Jan 7, 2020
@karelzak
tests: mark mdadm tests as TS_KNOWN_FAIL
4ae96cf 
It seems recent changes in mdadm make things a little bit unstable for
some users. Let's mark the tests temporary as TS_KNOWN_FAIL until we
get more information.

Addresses: #906
Signed-off-by: Karel Zak <kzak@redhat.com>
@karelzak
Copy link
Collaborator

karelzak commented Jan 7, 2020

I have doubts it's about kernel version only. it seems more complicated, for example it works on Fedora with kernel 5.5

# ./ts/blkid/md-raid0-whole; ./ts/blkid/md-raid1-part; ./ts/blkid/md-raid1-whole; ./ts/fdisk/align-512-4K-md
        blkid: MD raid0 (whole-disks)         ... OK
        blkid: MD raid1 (last partition)      ... OK
        blkid: MD raid1 (whole-disks)         ... OK
        fdisk: align 512/4K +MD               ... OK

# uname -a
Linux 10.255.255.10 5.5.0-0.rc4.git2.1.fc32.x86_64 #1 SMP Fri Jan 3 20:19:07 UTC 2020 x86_64 x86_64 x86_64 GNU/Linux
# mdadm -V
mdadm - v4.1-rc2 - 2018-08-02

Two weeks ago I fixed issue with raid members size related to the default layout setting, commit 653d1c8 (https://bugzilla.redhat.com/show_bug.cgi?id=1784534) ... not sure, maybe it's related to the issue too.

Anyway, I have a temporary marked the tests as TS_KNOWN_FAIL, I need to test it with the unstable mdadm to debug it more carefully. We'll see. I'm going to keep this issue open for now.

@mator
Copy link
Contributor Author

mator commented Jun 8, 2020 •
edited
Loading 

util-linux/tests# git desc
v2.35-380-gac26b19ba

util-linux/tests# uname -a
Linux ttip 5.7.0-12703-gaf7b4801030c #142 SMP Mon Jun 8 11:34:57 MSK 2020 sparc64 GNU/Linux

# mdadm -V
mdadm - v4.1 - 2018-10-01

util-linux/tests# ./ts/blkid/md-raid0-whole; ./ts/blkid/md-raid1-part; ./ts/blkid/md-raid1-whole; ./ts/fdisk/align-512-4K-md
        blkid: MD raid0 (whole-disks)         ... KNOWN FAILED (blkid/md-raid0-whole)
        blkid: MD raid1 (last partition)      ... OK
        blkid: MD raid1 (whole-disks)         ... OK
        fdisk: align 512/4K +MD               ... KNOWN FAILED (fdisk/align-512-4K-md)

@karelzak karelzak removed the ASSIGNED label Dec 6, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@mator @karelzak