Thank you both for your reply,
The "DBUS :WARN :Health status is unhealthy" is weird because the volume is not having any workload, it's just mounted by ESXi servers and the vms are shutdown, also all bricks are SSDs.
You mentioned that it might be related to requests queue being full, where can i check the current queue status ?
--
Respectfully
Mahdi A. Mahdi
Respectfully
Mahdi A. Mahdi
From: Nithya Balachandran <nbalacha@xxxxxxxxxx>
Sent: Thursday, March 2, 2017 8:32:52 AM
To: Soumya Koduri
Cc: Mahdi Adnan; gluster-users@xxxxxxxxxxx; Krutika Dhananjay; Frank Filz
Subject: Re: nfs-ganesha logs
Sent: Thursday, March 2, 2017 8:32:52 AM
To: Soumya Koduri
Cc: Mahdi Adnan; gluster-users@xxxxxxxxxxx; Krutika Dhananjay; Frank Filz
Subject: Re: nfs-ganesha logs
On 1 March 2017 at 18:25, Soumya Koduri
<skoduri@xxxxxxxxxx> wrote:
I am not sure if there are any outstanding issues with exposing shard volume via gfapi. CCin Krutika.
On 02/28/2017 01:29 PM, Mahdi Adnan wrote:
Hi,
We have a Gluster volume hosting VMs for ESXi exported via Ganesha.
Im getting the following messages in ganesha-gfapi.log and ganesha.log
=====
[2017-02-28 07:44:55.194621] E [MSGID: 109040]
[dht-helper.c:1198:dht_migration_complete_check_task] 0-vmware2-dht:
<gfid:ec846aeb-50f9-4b39-b0c9-24a8b833afe6>: failed to lookup the file
on vmware2-dht [Stale file handle]
This "Stale file handle" error suggests that the file may have just got removed at the back-end. Probably someone more familiar with dht (cc'ed Nithya) can confirm if there are other possibilities.
That is one possibility. In case a FOP returns an ENOENT/ESTALE because the file was deleted before it could go through, DHT checks to see if the file was migrated to another brick. However, as the file is no longer present on the volume, you will see
the dht_migration_complete_check_task message above.
You might want to check if the file in question still exists. There should also be messages in the log indicating which fop has failed.
Do you happen to have core? If yes, could you please check the bt. Below messages are just heartbeat warnings typically thrown when the outstanding request queue is above certain bench mark and nfs-ganesha server is taking a while to process them. Also you seem to be using nfs-ganesha 2.3.x version. Its not being actively maintained. There are many improvements and fixes done in nfs-ganesha 2.4.x. I suggest to try out that version if possible.
[2017-02-28 07:44:55.194660] E [MSGID: 133014]
[shard.c:1129:shard_common_stat_cbk] 0-vmware2-shard: stat failed:
ec846aeb-50f9-4b39-b0c9-24a8b833afe6 [Stale file handle]
[2017-02-28 07:44:55.207154] W [MSGID: 108008]
[afr-read-txn.c:228:afr_read_txn] 0-vmware2-replicate-5: Unreadable
subvolume -1 found with event generation 8 for gfid
4a50127e-4403-49a5-9886-80541a76299c. (Possible split-brain)
[2017-02-28 07:44:55.209205] E [MSGID: 109040]
[dht-helper.c:1198:dht_migration_complete_check_task] 0-vmware2-dht:
<gfid:4a50127e-4403-49a5-9886-80541a76299c>: failed to lookup the file
on vmware2-dht [Stale file handle]
[2017-02-28 07:44:55.209265] E [MSGID: 133014]
[shard.c:1129:shard_common_stat_cbk] 0-vmware2-shard: stat failed:
4a50127e-4403-49a5-9886-80541a76299c [Stale file handle]
[2017-02-28 07:44:55.212556] W [MSGID: 108008]
[afr-read-txn.c:228:afr_read_txn] 0-vmware2-replicate-4: Unreadable
subvolume -1 found with event generation 2 for gfid
cec80035-1f51-434a-9dbf-8bcdd5f4a8f7. (Possible split-brain)
[2017-02-28 07:44:55.214702] E [MSGID: 109040]
[dht-helper.c:1198:dht_migration_complete_check_task] 0-vmware2-dht:
<gfid:cec80035-1f51-434a-9dbf-8bcdd5f4a8f7>: failed to lookup the file
on vmware2-dht [Stale file handle]
[2017-02-28 07:44:55.214741] E [MSGID: 133014]
[shard.c:1129:shard_common_stat_cbk] 0-vmware2-shard: stat failed:
cec80035-1f51-434a-9dbf-8bcdd5f4a8f7 [Stale file handle]
[2017-02-28 07:44:55.259729] I [MSGID: 108031]
[afr-common.c:2154:afr_local_discovery_cbk] 0-vmware2-replicate-0:
selecting local read_child vmware2-client-0
[2017-02-28 07:44:55.259937] I [MSGID: 108031]
[afr-common.c:2154:afr_local_discovery_cbk] 0-vmware2-replicate-4:
selecting local read_child vmware2-client-8
=====
28/02/2017 06:27:54 : epoch 58b05af4 : gluster01 :
ganesha.nfsd-2015[dbus_heartbeat] dbus_heartbeat_cb :DBUS :WARN :Health
status is unhealthy. Not sending heartbeat
28/02/2017 06:33:36 : epoch 58b05af4 : gluster01 :
ganesha.nfsd-2015[work-9] cache_inode_avl_qp_insert :INODE :CRIT
:cache_inode_avl_qp_insert_s: name conflict (access, access)
=====
The volume is hosting a few VMs without any noticeable workload, and all
bricks are SSDs.
Im censored about the logs messages because i have another cluster and
ganesha keeps on crashing every few days with the following message
spamming the log:
AFAIK, there are no tools integrated with nfs-ganesha which monitors and displays IO statistics. Request Frank & others to comment.
>
28/02/2017 08:02:45 : epoch 58b1e2f5 : gfs01 :
ganesha.nfsd-31929[dbus_heartbeat] dbus_heartbeat_cb :DBUS :WARN :Health
status is unhealthy. Not sending heartbeat
28/02/2017 08:41:08 : epoch 58b1e2f5 : gfs01 :
ganesha.nfsd-31929[dbus_heartbeat] dbus_heartbeat_cb :DBUS :WARN :Health
status is unhealthy. Not sending heartbeat
28/02/2017 08:48:38 : epoch 58b1e2f5 : gfs01 :
ganesha.nfsd-31929[dbus_heartbeat] dbus_heartbeat_cb :DBUS :WARN :Health
status is unhealthy. Not sending heartbeat
28/02/2017 08:48:52 : epoch 58b1e2f5 : gfs01 :
ganesha.nfsd-31929[dbus_heartbeat] dbus_heartbeat_cb :DBUS :WARN :Health
status is unhealthy. Not sending heartbeat
28/02/2017 09:16:27 : epoch 58b1e2f5 : gfs01 :
ganesha.nfsd-31929[dbus_heartbeat] dbus_heartbeat_cb :DBUS :WARN :Health
status is unhealthy. Not sending heartbeat
28/02/2017 09:46:54 : epoch 58b1e2f5 : gfs01 :
ganesha.nfsd-31929[dbus_heartbeat] dbus_heartbeat_cb :DBUS :WARN :Health
status is unhealthy. Not sending heartbeat
28/02/2017 09:50:02 : epoch 58b1e2f5 : gfs01 :
ganesha.nfsd-31929[dbus_heartbeat] dbus_heartbeat_cb :DBUS :WARN :Health
status is unhealthy. Not sending heartbeat
28/02/2017 09:57:03 : epoch 58b1e2f5 : gfs01 :
ganesha.nfsd-31929[dbus_heartbeat] dbus_heartbeat_cb :DBUS :WARN :Health
status is unhealthy. Not sending heartbeat
28/02/2017 09:57:14 : epoch 58b1e2f5 : gfs01 :
ganesha.nfsd-31929[dbus_heartbeat] dbus_heartbeat_cb :DBUS :WARN :Health
status is unhealthy. Not sending heartbeat
28/02/2017 10:48:41 : epoch 58b1e2f5 : gfs01 :
ganesha.nfsd-31929[dbus_heartbeat] dbus_heartbeat_cb :DBUS :WARN :Health
status is unhealthy. Not sending heartbeat
SSDs volumes are running Gluster 3.8.9 and Ganesha V2.3.3 and the other
cluster is running Gluster 3.7.19 and Ganesha V2.3.0.
also, how can i get IO statics from Ganesha ?
Thanks,
Soumya
I appreciate your help.
--
Respectfully*
**Mahdi A. Mahdi*
_______________________________________________
Gluster-users mailing list
Gluster-users@xxxxxxxxxxx
http://lists.gluster.org/mailman/listinfo/gluster-users
_______________________________________________ Gluster-users mailing list Gluster-users@xxxxxxxxxxx http://lists.gluster.org/mailman/listinfo/gluster-users
