Description of problem:
Today, mkfs.gfs2 enforces a limit of 16 bytes for cluster name
(and 16 bytes for file system name) in the superblock, even
though there's room for 64 bytes (total) for sb_locktable.
The trouble is, pacemaker / corosync allow much longer cluster
names (not really limited). The dlm_controld seems to have a
64-byte limit*. Customers have complained that they create a
cluster, get it all running, then when they try to make the GFS2
file systems, mkfs.gfs2 rejects the cluster name for being too
long. However, renaming the cluster is not very easy to do.
I'd like to see us double the limit from 16:16 to 32:31 (need
to leave room for the colon).
Version-Release number of selected component (if applicable):
Steps to Reproduce:
# mkfs.gfs2 -O -p lock_dlm -j4 -t "financial_cluster:intec34p" /dev/mpathc/intec34p
locktable error: cluster name too long
libdlm/libdlm.h:#define DLM_LOCKSPACE_LEN 64
Created attachment 1018051 [details]
Patch posted upsream
This patch increases the allowed cluster name to 32 bytes along
with increasing the allowed the file system name to 30 bytes.
Verified in gfs2-utils-3.1.8-4:
[root@host-050 ~]# rpm -q gfs2-utils
[root@host-050 ~]# mkfs.gfs2 -O -p lock_dlm -j4 -t "financial_cluster:intec34p" /dev/sda1
This will destroy any data on /dev/sda1
Block size: 4096
Device size: 200.00 GB (52428544 blocks)
Filesystem size: 200.00 GB (52428541 blocks)
Resource groups: 802
Locking protocol: "lock_dlm"
Lock table: "financial_cluster:intec34p"
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.
For information on the advisory, and where to find the updated
files, follow the link below.
If the solution does not work for you, open a new bug report.