Bug 1054678 - nfs-ganesha: script does not disable glusterfs-nfs in case gluterfs-nfs is already not running on the present node of the rhs cluster
Summary: nfs-ganesha: script does not disable glusterfs-nfs in case gluterfs-nfs is al...
Keywords:
Status: CLOSED NOTABUG
Alias: None
Product: Red Hat Gluster Storage
Classification: Red Hat Storage
Component: nfs-ganesha
Version: 2.1
Hardware: x86_64
OS: Linux
medium
medium
Target Milestone: ---
: ---
Assignee: Anand Subramanian
QA Contact: Saurabh
URL:
Whiteboard:
Depends On:
Blocks: 1035040
TreeView+ depends on / blocked
 
Reported: 2014-01-17 09:23 UTC by Saurabh
Modified: 2016-01-19 06:16 UTC (History)
9 users (show)

Fixed In Version: nfs-ganesha-2.0.0.2-4.el6rhs
Doc Type: Known Issue
Doc Text:
Before starting nfs-ganesha, gluster-nfs should be disabled by setting nfs.disable option to ON. The rhs-nfs_ganesha.sh script checks only if gluster-nfs process is running on the host and if gluster-nfs is down for some reason , it does not set the nfs.disable option to ON. This might result in two issues: 1.nfs-ganesha starts on the host specified as input to the rhs-nfs_ganesha.sh script. But, gluster-nfs would still be active on the nodes other than the node running nfs-ganesha if the volume has bricks across multiple nodes. Workaround: Set nfs.disable option to ON by using the following command on any of the nodes containing the RHS volume in question, gluster volume set [volume-name] nfs.disable ON 2.nfs-ganesha fails to start and the script displays the message, "NFS ports still appear to be in use, please try again." Workaround: Follow the listed steps to troubleshoot this issue, a.Check the log file to see if the failure is due to ports still being in use. b.Check rpcinfo -p to see if NFS ports are still in use. c.Remove the independent entries for nfs,mountd,nlockmgr and nfs_acl using rpcinfo -d command. d.Execute the following command, gluster volume set [volume-name] nfs.disable ON e.Start NFS-ganesha using the script.
Clone Of:
Environment:
Last Closed: 2014-07-07 12:24:34 UTC
Embargoed:


Attachments (Terms of Use)

Description Saurabh 2014-01-17 09:23:40 UTC
Description of problem:
rhs-nfs_ganesha.sh script is used to invoke the nfs-ganesha process.
Before invoking the process it asks for disabling the glusterfs-nfs. 
The same behavior is not seen if the glusterfs-nfs is not running on the present node, the behvaior related to quering about disabling the glustefs-nfs. Although the nfs-ganesha is invoked successfully.

This is a problem because on other nodes of the cluster, the glusterfs nfs process will still be running. 

In this scenario, lets say if we have four RHS nodes [1, 2, 3, 4]
and on node 1 somehow the glusterfs-nfs is killed or not running at the time we try to bring the nfs-ganesha process on this node. 
So, nfs-ganesha will be invoked without disabling nfs on the cluster, as query related to disabling will not be done. Therefore, on node 1 we will have nfs-ganesha and on other nodes 2, 3, 4 we will have glusterfs-nfs running. 

May be a confusing stage for the storage-system-admin.

Version-Release number of selected component (if applicable):
glusterfs-3.4.0.57rhs-1.el6rhs.x86_64
nfs-ganesha-2.0.0.1-5.el6rhs.x86_64

How reproducible:
always

Additional info:

Comment 2 Meghana 2014-02-07 11:05:48 UTC
The issue is fixed in the last rhs nfs-ganesha build (nfs-ganesha-2.0.0.2-4.el6rhs). Please try the above bld and verify the fix.

Comment 3 Gowrishankar Rajaiyan 2014-02-07 12:47:47 UTC
(In reply to Meghana from comment #2)
> The issue is fixed in the last rhs nfs-ganesha build
> (nfs-ganesha-2.0.0.2-4.el6rhs). Please try the above bld and verify the fix.

It would be awesome if you can point out the patch URL for this fix.

- shanks

Comment 4 Saurabh 2014-02-17 09:26:51 UTC
well, seems the issue is still not fixed, the other nodes are still having glusterfs-nfs running.

[root@rhsauto004 ~]# ps -eaf | grep nfs
root      3574     1  6 14:44 ?        00:00:14 /usr/sbin/glusterfs -s localhost --volfile-id gluster/nfs -p /var/lib/glusterd/nfs/run/nfs.pid -l /var/log/glusterfs/nfs.log -S /var/run/c66ee1ad00917f211581417cb71f0e9b.socket
root      3666  3122  0 14:48 pts/0    00:00:00 grep nfs
[root@rhsauto004 ~]# kill -9 3574
[root@rhsauto004 ~]# 
[root@rhsauto004 ~]# 
[root@rhsauto004 ~]# time bash /usr/lib/glusterfs/.unsupported/rhs-nfs_ganesha.sh -l /root/ganeshalog/1.log -v dist-rep -i 10.70.36.234 -f /usr/lib/glusterfs/.unsupported/nfs-ganesha.conf -c start -n 3

NFS-ganesha server started.
============Volume exports============
Export list for 10.70.36.234:
/dist-rep *,*
======================================

real	0m6.769s
user	0m0.478s
sys	0m0.305s
[root@rhsauto004 ~]# gluster volume info
 
Volume Name: dist-rep
Type: Distributed-Replicate
Volume ID: 414a3d23-9e22-436e-a527-984a003657a8
Status: Started
Number of Bricks: 6 x 2 = 12
Transport-type: tcp
Bricks:
Brick1: rhsauto004.lab.eng.blr.redhat.com:/rhs/brick1/d1r1
Brick2: rhsauto012.lab.eng.blr.redhat.com:/rhs/brick1/d1r2
Brick3: rhsauto015.lab.eng.blr.redhat.com:/rhs/brick1/d2r1
Brick4: rhsauto016.lab.eng.blr.redhat.com:/rhs/brick1/d2r2
Brick5: rhsauto004.lab.eng.blr.redhat.com:/rhs/brick1/d3r1
Brick6: rhsauto012.lab.eng.blr.redhat.com:/rhs/brick1/d3r2
Brick7: rhsauto015.lab.eng.blr.redhat.com:/rhs/brick1/d4r1
Brick8: rhsauto016.lab.eng.blr.redhat.com:/rhs/brick1/d4r2
Brick9: rhsauto004.lab.eng.blr.redhat.com:/rhs/brick1/d5r1
Brick10: rhsauto012.lab.eng.blr.redhat.com:/rhs/brick1/d5r2
Brick11: rhsauto015.lab.eng.blr.redhat.com:/rhs/brick1/d6r1
Brick12: rhsauto016.lab.eng.blr.redhat.com:/rhs/brick1/d6r2
Options Reconfigured:
nfs.disable: off

Comment 6 Shalaka 2014-02-18 08:42:16 UTC
Please add doc text for this known issue.

Comment 7 Shalaka 2014-02-19 06:29:25 UTC
Please review the edited doc text and sign off.

Comment 9 Soumya Koduri 2014-07-07 12:24:34 UTC
This is no longer an issue and importantly not applicable to Denali RHS 3.0 Ganesha. The script rhs_nfs-ganehsa.sh is deprecated.


Note You need to log in before you can comment on or make changes to this bug.