Open figure002 opened 3 months ago
This issue is discussed many times in the mailing list (https://lists.gluster.org/pipermail/gluster-users/2015-March/021333.html) and other issues. We need to come up with the proper solution to this use case.
Issue:
[2024-02-13 16:12:31.713359 +0000] I [fuse-bridge.c:5927:fuse_graph_sync] 0-fuse: switched to graph 0
[2024-02-13 16:12:31.715049 +0000] W [MSGID: 108027] [afr-common.c:2889:afr_attempt_readsubvol_set] 0-vol1-replicate-0: no read subvols for /
[2024-02-13 16:12:31.715113 +0000] E [fuse-bridge.c:5365:fuse_first_lookup] 0-fuse: first lookup on root failed (Transport endpoint is not connected)
[2024-02-13 16:12:31.715982 +0000] W [fuse-bridge.c:1403:fuse_attr_cbk] 0-glusterfs-fuse: 17: LOOKUP() / => -1 (Transport endpoint is not connected)
[2024-02-13 16:12:31.720745 +0000] W [fuse-bridge.c:1403:fuse_attr_cbk] 0-glusterfs-fuse: 18: LOOKUP() / => -1 (Transport endpoint is not connected)
[2024-02-13 16:12:31.721302 +0000] I [fuse-bridge.c:6308:fuse_thread_proc] 0-fuse: initiating unmount of /mnt/gluster
The message "W [MSGID: 108027] [afr-common.c:2889:afr_attempt_readsubvol_set] 0-vol1-replicate-0: no read subvols for /" repeated 2 times between [2024-02-13 16:12:31.715049 +0000] and [2024-02-13 16:12:31.720735 +0000]
[2024-02-13 16:12:31.722241 +0000] W [glusterfsd.c:1427:cleanup_and_exit] (-->/lib/x86_64-linux-gnu/libpthread.so.0(+0x7ea7) [0x7fe6c2524ea7] -->/usr/sbin/glusterfs(+0x12725) [0x56075d8d3725] -->/usr/sbin/glusterfs(cleanup_and_exit+0x57) [0x56075d8cbd77] ) 0-: received signum (15), shutting down
[2024-02-13 16:12:31.722263 +0000] I [fuse-bridge.c:7051:fini] 0-fuse: Unmounting '/mnt/gluster'.
[2024-02-13 16:12:31.722273 +0000] I [fuse-bridge.c:7055:fini] 0-fuse: Closing fuse connection to '/mnt/gluster'.
Expected:
ENOTCONN
during IO. @amarts @pranithk what you guys think?
I think we can do it after make the option transport.socket.ignore-enoent to configurable. As of now the option is hard coded and use by glusterd, it call connect in a loop at max 5 times after 1s sleep even connect received ENOENT.
so many glusterd nodes, so we have sleep 10 sec,and retry,until all subvolume is up or 20 times limit。 i don't known the common solution, but we use this solution and work fine.
Description of problem:
I have a GlusterFS cluster with 3 nodes. All nodes serve as both server and client, as the servers are running Docker containers that use the GlusterFS volume.
My
/etc/fstab
:But the volume fails to mount at boot. But if I login via SSH and run
mount -a
, the mount works just fine.The exact command to reproduce the issue:
Expected results:
The volume should automatically mount at boot.
Mandatory info:
- The output of the
gluster volume info
command:- The output of the
gluster volume status
command:- The output of the
gluster volume heal
command:- Provide logs present on following locations of client and server nodes -
/var/log/syslog
/var/log/glusterfs/glusterd.log
/var/log/glusterfs/mnt-gluster.log
/var/log/glusterfs/glustershd.log
Additional info:
In an attempt to fix the above issue, I created some systemd drop-in files to ensure that the volume isn't mounted until the GlusterFS service is started:
I can tell from the timestamps that the mount isn't done until GlusterFS is started, but it still fails to mount. The above logs are with the drop-in files in place.
- The operating system / glusterfs version:
Debian GNU/Linux 11 (bullseye)
glusterfs-server: Installed: 11.1-1
glusterfs-client: Installed: 11.1-1