[lustre-discuss] Correct use of "lctl get_param health_check"
Thomas Roth
dibbegucker at googlemail.com
Sun Jun 21 11:33:57 UTC 2026
Hi Dominic,
it us a useful parameter to include in some monitoring/alerting.
We've had unhealthy OSTs from overloading the servers, for example. Ideally, your monitoring would alert you _before_ an OST becomes unhealthy, but if
not, at least then you have to intervene. (Although Lustre goes on, of course - can go on even with an unhealthy MDT ... until your users start
protesting).
If a server does not have Lustre mounted, there is nothing there that could be unhealthy - therefore you shouldn't get a health warning then.
Of course, you need another monitoring to check if all your wanted targets are actually mounted.
Cheers
Thomas
On 6/11/26 3:21 PM, Dominic Sloan-Murphy via lustre-discuss wrote:
> Hi all,
>
> Would you be able to advise on the correct use of the "lctl get_param health_check" command from https://wiki.lustre.org/Lustre_Health_Checks
> <https://wiki.lustre.org/Lustre_Health_Checks> and under which circumstances it reports errors?
>
> I've found it always returns "healthy", even when I intentionally break my test setup. For example, removing Lustre mounts:
>
> root at lustre-build:~# mount -t lustre mytestpool/mgsmdt /mnt/mgs_mdt
> root at lustre-build:~# mount -t lustre ostpool/ost1 /mnt/ost
> root at lustre-build:~# lctl get_param health_check
> health_check=healthy
> root at lustre-build:~# umount /mnt/ost
> root at lustre-build:~# lctl get_param health_check
> health_check=healthy
> root at lustre-build:~# umount /mnt/mgs_mdt
> root at lustre-build:~# lctl get_param health_check
> health_check=healthy
>
> My setup is a single Ubuntu 24.04 VM with Lustre 2.17 built from source. The Lustre file system was created with:
>
> root at lustre-build:~# truncate -s 5G /root/mgs_mdt.img
> root at lustre-build:~# truncate -s 10G /root/ost.img
> root at lustre-build:~# mkfs.lustre --mgs --mdt --backfstype=zfs --fsname=mytestfs --index=0 mytestpool/mgsmdt /root/mgs_mdt.img
> root at lustre-build:~# MGS_IP=$(hostname -I | awk '{print $1}')
> root at lustre-build:~# mkfs.lustre --ost --backfstype=zfs --fsname=mytestfs --mgsnode=${MGS_IP}@tcp0 --index=0 ostpool/ost1 /root/ost.img
>
> Regards,
> Dominic
>
--
--------------------------------------------------------------------
Thomas Roth
Department: IT
GSI Helmholtzzentrum für Schwerionenforschung GmbH
Planckstraße 1, 64291 Darmstadt, Germany, www.gsi.de
Commercial Register / Handelsregister: Amtsgericht Darmstadt, HRB 1528
Managing Directors / Geschäftsführung:
Prof. Dr. Thomas Nilsson, Dr. Katharina Stummeyer, Jörg Blaurock
Chairman of the Supervisory Board / Vorsitzende des GSI-Aufsichtsrats:
State Secretary / Ministerialrätin Dr. Andrea Fischer
More information about the lustre-discuss
mailing list