Bug 961681

Summary: nfs: showmount fails to give the information
Product: [Red Hat Storage] Red Hat Gluster Storage Reporter: Saurabh <saujain>
Component: glusterdAssignee: rjoseph
Status: CLOSED CURRENTRELEASE QA Contact: Saurabh <saujain>
Severity: high Docs Contact:
Priority: high    
Version: 2.1CC: mzywusko, rhs-bugs, vbellur
Target Milestone: ---   
Target Release: ---   
Hardware: x86_64   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2013-06-14 12:18:23 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:

Description Saurabh 2013-05-10 09:20:40 UTC 
Description of problem:
showmount <ip or hostname of server>
gives the response with the ip of the clients that have nfs mount from the server.

Version-Release number of selected component (if applicable):
glusterfs-3.4.0.5rhs-1.el6rhs.x86_64

How reproducible:
alwyas, on this build, not earlier

Steps to Reproduce:
1. create a volume , start a volume using rhs nodes (a b c d)
2. nfs mount the volume from node "a" on clients (c1 c2)
3. showmount <ip of "a"> on any machine
  
Actual results:
should display the list of clients that mounts,

[root@bigbend1 ~]# showmount 10.70.37.155
Hosts on 10.70.37.155:


Expected results:
in similar fashion as kernel nfs mount does it,
[root@rhel6 ~]# showmount 10.70.34.114
Hosts on 10.70.34.114:
10.10.49.184
10.10.49.32
10.10.49.38
10.10.50.112
10.10.51.149
10.10.52.106
10.10.52.127
10.10.53.171
10.10.53.79
10.10.61.144
10.16.70.11


Additional info:
Comment 2 rjoseph 2013-05-21 10:31:40 UTC 
Tested on the latest code and glusterfs-3.4.0.8rhs-1.el6rhs.x86_64 both of them shows the correct behaviour.
Bug is not reproduced.

Apart from mounting any other operation should I do to reproduce the problem?
Comment 3 rjoseph 2013-05-22 07:05:57 UTC 
Working fine on latest build.
Comment 4 rjoseph 2013-06-05 10:16:01 UTC 
In the following test I am able reproduce the bug.

1. create a volume , start a volume using rhs nodes (a b c d)
2. nfs mount the volume from node "a" on clients (c1 c2)
3. showmount <ip of "a"> on any machine 
      --> This works fine shows IP of c1 and c2

Let's say on c1 nfs is mounted on /mnt/c1 and on c2 /mnt/c2
4. cd /mnt/c1 on c1
5. umount /mnt/c1 on c1 
      ---> Here unmount will with "device is busy"
6. check showmount <ip pf "a">   
      ---> c1 is removed from the entry even though it is still mounted.


The problem is that NFS client is sending unmount even if the client fails with "device is busy". 
The same behaviour can be seen in NFS kernel implementation.