[HTCondor-users] BUG: LVM_* related config does not handle the "-" in names well

[Alexandr Mikula <mikula@xxxxxx>]

Hi,
the LVM_* related config does not handle the "-" (dash) in the volume
group name very well. (I guess this also applies the thin pool name)
config:
"""
LVM_VOLUME_GROUP_NAME = mff-mikan05vg
LVM_THINPOOL_NAME = htc_thinpool
LVM_USE_THIN_PROVISIONING = True
"""

Part of lsblk:
    ââmff--mikan05vg-htc_thinpool_tmeta   253:2    0   968M  0 lvm   
    â ââmff--mikan05vg-htc_thinpool-tpool 253:4    0 964.6G  0 lvm   
    â   ââmff--mikan05vg-htc_thinpool     253:5    0 964.6G  1 lvm   
    ââmff--mikan05vg-htc_thinpool_tdata   253:3    0 964.6G  0 lvm   
    â ââmff--mikan05vg-htc_thinpool-tpool 253:4    0 964.6G  0 lvm   
    â   ââmff--mikan05vg-htc_thinpool     253:5    0 964.6G  1 lvm  

The problem is the condor does not "add" the extra "-" to the
LVM_VOLUME_GROUP_NAME when it tries to create the filesystem and
creates following error:

"""
10/14/25 10:28:01 (pid:634001) setting the orig job iwd in starter
10/14/25 10:28:01 (pid:634001) Failed to setup LVM filesystem for job:
VolumeManager:13:Failed to create new ext4 filesystem for job (exit
status 256): Warning: label too long; will be truncated to
'slot1_1+587183_4'

mke2fs 1.46.5 (30-Dec-2021)
The file /dev/mapper/mff-mikan05vg-slot1_1+587183_458 does not exist
and no size was specified.


10/14/25 10:28:02 (pid:634001) Failed to initialize
JobInfoCommunicator, aborting
"""

it fails and creates the zombie process for startd.

Also it would be nice to limit the label size on the condor side not to
produce the label warning.

Cheers
AM
-- 
Alexandr Mikula
OddÄlenà sÃÅovÃnà a vÃpoÄetnà techniky & VÃpoÄetnà stÅedisko 
FyzikÃlnà Ãstav Akademie vÄd Äeskà republiky, v. v. i.
Institute of Physics of the Czech Academy of Sciences

Attachment: smime.p7s
Description: S/MIME cryptographic signature