答复: some bricks didn't come up after a server reboot

[likun <kun.li@xxxxxxxxxxx>]

Yes, I could ensure that no other daemon used this port at that time.

I have encountered this situation several times, some need multiple restarts to bring all bricks back. Before restart, I killed all the daemons, and used netstat to ensure the ports begin with 4915  were closed , and also 24007 .

 

Likun

 

On Mon, Jan 16, 2017 at 10:19 AM, likun <kun.li@xxxxxxxxxxx> wrote:

Hi, everybody.

 

I have glusterfs 3.8.5 setuped in a kubernetes enveironment.

 

Recently, one of my glusterfs server rebooted, glusterfs service didn’t come up successfully.

 

Out of totally 6 bricks, 3 came up, 3 didn’t.

 

Here is the logs of the brick that have failed:

 

[2016-12-29 03:46:50.240032] I [MSGID: 100030] [glusterfsd.c:2454:main] 0-/usr/sbin/glusterfsd: Started running /usr/sbin/glusterfsd version 3.8.5 (args: /usr/sbin/glusterfsd -s 10.32.3.23 --volfile-id gvol0.10.32.3.23.mnt-brick1-vol -p /var/lib/glusterd/vols/gvol0/run/10.32.3.23-mnt-brick1-vol.pid -S /var/run/gluster/08da045f3e66eefc50c0ff9a035c6794.socket --brick-name /mnt/brick1/vol -l /var/log/glusterfs/bricks/mnt-brick1-vol.log --xlator-option *-posix.glusterd-uuid=58c3b462-a4b6-4655-b2ac-d0502e278e03 --brick-port 49152 --xlator-option gvol0-server.listen-port=49152)

 

So 49152 port was allocated for this brick.
 

[2016-12-29 03:46:50.258772] I [MSGID: 101190] [event-epoll.c:628:event_dispatch_epoll_worker] 0-epoll: Started thread with index 1

[2016-12-29 03:47:08.153575] I [MSGID: 101173] [graph.c:269:gf_add_cmdline_options] 0-gvol0-server: adding option 'listen-port' for volume 'gvol0-server' with value '49152'

[2016-12-29 03:47:08.153613] I [MSGID: 101173] [graph.c:269:gf_add_cmdline_options] 0-gvol0-posix: adding option 'glusterd-uuid' for volume 'gvol0-posix' with value '58c3b462-a4b6-4655-b2ac-d0502e278e03'

[2016-12-29 03:47:08.153777] I [MSGID: 115034] [server.c:398:_check_for_auth_option] 0-gvol0-decompounder: skip format check for non-addr auth option auth.login./mnt/brick1/vol.allow

[2016-12-29 03:47:08.153785] I [MSGID: 115034] [server.c:398:_check_for_auth_option] 0-gvol0-decompounder: skip format check for non-addr auth option auth.login.94bedfd1-619d-402a-9826-67dab7600f43.password

[2016-12-29 03:47:08.153895] I [MSGID: 101190] [event-epoll.c:628:event_dispatch_epoll_worker] 0-epoll: Started thread with index 2

[2016-12-29 03:47:08.159776] I [rpcsvc.c:2214:rpcsvc_set_outstanding_rpc_limit] 0-rpc-service: Configured rpc.outstanding-rpc-limit with value 64

[2016-12-29 03:47:08.159829] W [MSGID: 101002] [options.c:954:xl_opt_validate] 0-gvol0-server: option 'listen-port' is deprecated, preferred is 'transport.socket.listen-port', continuing with correction

[2016-12-29 03:47:08.159893] E [socket.c:793:__socket_server_bind] 0-tcp.gvol0-server: binding to  failed: Address already in use

[2016-12-29 03:47:08.159899] E [socket.c:796:__socket_server_bind] 0-tcp.gvol0-server: Port is already in use

 

What does 'netstat -nap | grep 49152' says at that time? Are we sure no other application has accidentally raced to consume the same port  ?

[2016-12-29 03:47:08.159907] W [rpcsvc.c:1645:rpcsvc_create_listener] 0-rpc-service: listening on transport failed

[2016-12-29 03:47:08.159913] W [MSGID: 115045] [server.c:1061:init] 0-gvol0-server: creation of listener failed

[2016-12-29 03:47:08.159919] E [MSGID: 101019] [xlator.c:433:xlator_init] 0-gvol0-server: Initialization of volume 'gvol0-server' failed, review your volfile again

[2016-12-29 03:47:08.159924] E [MSGID: 101066] [graph.c:324:glusterfs_graph_init] 0-gvol0-server: initializing translator failed

[2016-12-29 03:47:08.159929] E [MSGID: 101176] [graph.c:673:glusterfs_graph_activate] 0-graph: init failed

[2016-12-29 03:47:08.160764] W [glusterfsd.c:1327:cleanup_and_exit] (-->/usr/sbin/glusterfsd(mgmt_getspec_cbk+0x3c1) [0x55ead22bee51] -->/usr/sbin/glusterfsd(glusterfs_process_volfp+0x172) [0x55ead22b95d2] -->/usr/sbin/glusterfsd(cleanup_and_exit+0x6b) [0x55ead22b8b4b] ) 0-: received signum (98), shutting down

 

What’s wrong with it ? “port is already in use” ? the system was just rebooted, it’s clear, no other daemon was using this port .

After restarted the glusterfs container, all six bricks came up. Sometime, two restart or more needed to take all the bricks back to work. Any idea ?

 

Likun

 

_______________________________________________
Gluster-users mailing list
Gluster-users@xxxxxxxxxxx
http://lists.gluster.org/mailman/listinfo/gluster-users

--

 

~ Atin (atinm)

 

_______________________________________________
Gluster-users mailing list
Gluster-users@xxxxxxxxxxx
http://lists.gluster.org/mailman/listinfo/gluster-users