NFSv4 Troubleshooting
- Issues with listing the exports:
- If
nfs4mgr list-exports
command returns the error,Error org.freedesktop.DBus.Error.ServiceUnknown: The name org.ganesha.nfsd was not provided by any .service files
, do the following to resolve the error:- Open the
/etc/dbus-1/system.conf
file. - Find and replace the following in the file:
- Find:
<deny send_destination="org.freedesktop.DBus" send_interface="org.freedesktop.DBus" send_member="UpdateActivationEnvironment"/>
- Replace with:
<!-- Allow anyone to talk to the message bus --> <allow send_destination="org.freedesktop.DBus"/> <!-- But disallow some specific bus services --> <allow send_interface="*"/> <allow receive_interface="*"/> <allow own="*"/>
- Find:
- Save and close the file.
- Restart the messagebus and NFSv4 service. To restart the messagebus service, run the following command:
service messagebus restart
To restart NFSv4 service, see Starting, Stopping, and Restarting HPE Ezmeral Data Fabric NFSv4.
- Open the
- Lock expires as a result of problems in node connection or node failover:
- If the lock expires as a result of problems in node connection or node failover, all IOs
from the application will fail with EIO message to prevent the file from getting
corrupted. You can tune the kernel to reclaim lost locks and not fail writes on lock lease
expiration. To tune the kernel, run the following command:
echo Y > /sys/module/nfs/parameters/recover_lost_locks
WARNING This kernel tuning this might result in data corruption.
- User ID mapping is not working when NFSv4 server and client are in two different sub-domains:
- If user ID (with FQDN user names) mapping is not working with the installed
libnfsidmap
library (for example,libnfsidmap-0.25-19
on Centos), do the following:- Download the latest
libnfsidmap
library for the OS.For CentOS:
- Download the library source with the patches.
- Apply the patches till
libnfsidmap-0.25-dns-resolved.patch
. - Build the new library.
- Stop NFSv4 service and replace the old library with the new library.
- Replace the old
nsswitch.so
file with the newnsswitch.so
file that is compiled. - Restart rpcidmapd service and NFSv4 service.
- Download the latest
- NFS Ganesha crashes when unmounting the cluster:
- NFS Ganesha crashes when unmounting the cluster as a non-root user. Download and install
the following Kerberos 16 package:
sssd-krb5-common-1.16.0-19.el7.x86_64
sssd-krb5-1.16.0-19.el7.x86_64
- Discover realm command fails:
- If the
realm discover
command fails with following error:realm: Couldn't connect to realm service: Error calling StartServiceByName for org.freedesktop.realmd: GDBus.Error:org.freedesktop.DBus.Error.TimedOut: Activation of org.freedesktop.realmd timed out
, reboot the node and add Active Directory information in the/etc/resolv.conf
again. For example, your entry in the/etc/resolv.conf
file should look similar to the following:# cat /etc/resolv.conf nameserver 10.10.111.41 domain nfs4ad.com search nfs4ad.com lab qa.lab scale.lab perf.lab ipmi.lab
- Unable to start the service:
- If you or the warden is unable to restart the NFS service, do the following:
- Review the warden logs (in
$MAPR_HOME/logs/warden.log
file) to determine when the NFSv4 Service Alarm was raised. - Review the following NFSv4 server logs in
$MAPR_HOME/logs/nfs4/nfs4server.log-0
and$MAPR_HOME/logs/nfs4/fsal.log-0
, and other logs under$MAPR_HOME/logs/nfs4
on the node where the service went down to determine the cause for the error.The following example logs show some common causes for NFSv4 service shutting down such as license not present or an issue in the configuration.
# tail -f /opt/mapr/logs/nfs4/fsal.log-0 2018-08-10 04:04:20,3058 FATAL FuseOps fs/client/fuse/cc/fuse_ops_ll.c:505 Thread: 1209 No license found. Shutting down 2018-08-10 04:04:34,6487 ERROR FuseAPI fc/fuse_api.cc:1384 Thread: 8877 Shmid to be used by fcdebug 1003847690, guts 0 2018-08-10 04:04:34,7749 ERROR Cidcache fc/cidcache.cc:5448 Thread: 8877 License not found. Shutting down 2018-08-10 04:04:34,7749 FATAL FuseOps fs/client/fuse/cc/fuse_ops_ll.c:505 Thread: 8877 No license found. Shutting down 2018-08-10 04:04:48,6729 ERROR FuseAPI fc/fuse_api.cc:1384 Thread: 15412 Shmid to be used by fcdebug 1005748236, guts 0 2018-08-10 04:04:48,7412 ERROR Cidcache fc/cidcache.cc:5448 Thread: 15412 License not found. Shutting down 2018-08-10 04:04:48,7413 FATAL FuseOps fs/client/fuse/cc/fuse_ops_ll.c:505 Thread: 15412 No license found. Shutting down
# tail -f /opt/mapr/logs/nfs4/nfs4server.log-0 10/08/2018 T05:58:06.410328-0700 8163[none] [main] 713 :export_commit_common :Exporting to NFSv4 but not Pseudo path defined 10/08/2018 T05:58:06.410338-0700 8163[none] [main] 2267 :fsal_put :FSAL MAPR now unused 10/08/2018 T05:58:06.410369-0700 8163[none] [main] 1443 :build_default_root :Export 0 (/) successfully created 10/08/2018 T05:58:06.410373-0700 8163[none] [main] 476 :main :No export entries found in configuration file !!! 10/08/2018 T05:58:06.410380-0700 8163[none] [main] 219 :config_errs_to_log :Config File (/opt/mapr/conf/nfs4server.conf:104): Syntax error in statement 10/08/2018 T05:58:06.410384-0700 8163[none] [main] 219 :config_errs_to_log :Config File (/opt/mapr/conf/nfs4server.conf:65): Unknown parameter (nfs_track_memory) 10/08/2018 T05:58:06.410387-0700 8163[none] [main] 219 :config_errs_to_log :Config File (/opt/mapr/conf/nfs4server.conf:68): Unknown parameter (mapr_log_debug_level) 10/08/2018 T05:58:06.410389-0700 8163[none] [main] 219 :config_errs_to_log :Config File (/opt/mapr/conf/nfs4server.conf:95): 1 validation errors in block EXPORT 10/08/2018 T05:58:06.410392-0700 8163[none] [main] 219 :config_errs_to_log :Config File (/opt/mapr/conf/nfs4server.conf:95): Errors processing block (EXPORT) 10/08/2018 T05:58:06.411681-0700 8163[none] [main] 1040 :cache_inode_lru_pkginit :Setting the system-imposed limit on FDs to 65536.
- Take corrective action to rectify the cause for the error.
- Review the warden logs (in