Bug 164875 - client autofs hangs intermittently while mounting nfs server
Summary: client autofs hangs intermittently while mounting nfs server
Keywords:
Status: CLOSED WONTFIX
Alias: None
Product: Red Hat Enterprise Linux 4
Classification: Red Hat
Component: kernel
Version: 4.0
Hardware: i386
OS: Linux
medium
medium
Target Milestone: ---
: ---
Assignee: Steve Dickson
QA Contact: Red Hat Kernel QE team
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2005-08-02 08:05 UTC by Rajdeep Sengupta
Modified: 2012-06-20 16:02 UTC (History)
3 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed: 2012-06-20 16:02:14 UTC
Target Upstream Version:
Embargoed:


Attachments (Terms of Use)

Description Rajdeep Sengupta 2005-08-02 08:05:04 UTC
From Bugzilla Helper:
User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1)

Description of problem:
We have a Sun NIS environment where around 200 redhat machines are there. Where some of the servers are Enterprise 3 while other are Enterprise 4. The client machines are mostly enteprise 3 or redhat 7.2. 
Till the nfs servers were Enterprise 3, it was working fine, though we use to face some issue with autofs unable to mount the nfs server. But once we restart the client autofs, the problem disapears. But recently when we upgraded some nfs server to Enterprise4, then most of the client autofs started behaving abnormal. Even restarting the the client autofs donot help much, only rebooting the client is the only solution.
The client autofs version are autofs-3.1.7-21 in 7.2 clients or autofs-4.1.3-67 or autofs-4.1.3-131

Version-Release number of selected component (if applicable):
nfs-utils-1.0.6-46

How reproducible:
Sometimes

Steps to Reproduce:
1.boot a redhat 7.2 machine with autofs-3.1.7-21 
2.boot the nfs 4 server (Enterprise 4)and exportfs the file system.
 3. try to mount the the filesystem thru autofs 

Actual Results:  the client hangs and the machine needs rebooting

Expected Results:  The client should smoothly mount the exported filesystem

Additional info:

Comment 1 Jeff Moyer 2005-08-02 12:13:34 UTC
This is a very vague problem description.

Does the problem occur on the RHEL 3 clients?  If so, then please provide the
information asked for in the "Filing bug reports" section of the following URL:
  http://people.redhat.com/jmoyer/

If you can only reproduce with your RH 7.2 clients, then please configure syslog
to capture debug messages.  You can do this like so:

o Add a line like the following to your /etc/syslog.conf:

*.*			/var/log/debug

o Restart syslogd (or send it a HUP signal).

Then, when the problem occurs, attach the /var/log/debug file to this bugzilla.

Comment 2 Rajdeep Sengupta 2005-08-03 09:00:41 UTC
I am getting similar error in both redhat 7.2 as well as EL 3 clients. My 
auto.master looks like 
/delsoft yp:auto.home -o soft,intr --debug
(I have added --debug on your suggestion)
Also initially soft was not there, but whenever I use to access the filesystem 
of EL4 server, it use to hang. SO I added the soft option to avoid hang.
Earlier when I use to do 
--------------------------------------------------------------------
file /delsoft/analyzer/OVL/3.5.0_ovl_2/cheetah/lib/libspyglassVE-static-
Linux2.a
It use to hang, but now due to the soft,intr it doesnot hang and it gives error

/delsoft/analyzer/OVL/3.5.0_ovl_2/cheetah/lib/libspyglassVE-static-Linux2.a: 
file: read failed (Input/output error).
----------------------------------------------------------------------
Just FYI, the /delsoft/analyzer is in a EL4 server.
ALso I donot get any error in /var/log/debug file, so I am not attaching the 
same.
One more thing, if I contineously do
-------------------------------------------------------------- 
file /delsoft/analyzer/OVL/3.5.0_ovl_2/cheetah/lib/libspyglassVE-static-
Linux2.a
-------------------------------------------------------------------
sometimes it gives Input/output error and sometimes it shows proper status
-------------------------------------------------
rajdeep@colorado [44] 
file /delsoft/analyzer/OVL/3.5.0_ovl_2/cheetah/lib/libspyglassVE-static-
Linux2.a 
/delsoft/analyzer/OVL/3.5.0_ovl_2/cheetah/lib/libspyglassVE-static-Linux2.a: 
file: read failed (Input/output error).
rajdeep@colorado [45] 
file /delsoft/analyzer/OVL/3.5.0_ovl_2/cheetah/lib/libspyglassVE-static-
Linux2.a
/delsoft/analyzer/OVL/3.5.0_ovl_2/cheetah/lib/libspyglassVE-static-Linux2.a: 
ELF 32-bit LSB relocatable, Intel 80386, version 1, not stripped
---------------------------------------------------------------
But normally if you access for the first time, it gives error while 2nd or 3rd 
time it doesnot give any error.
I have also noticed that if you wait for some time and again do file, it will 
gagain give error first time and then start working fine in 2nd or 3rd time.

The uname -a output of my 7.2 machine is
---------------------------------------------------- 
uname -a 
Linux colorado 2.4.7-10 #1 Thu Sep 6 17:27:27 EDT 2001 i686 unknown
-------------------------------------------------------------
while the the uname -a output of my EL 3 machine is 
--------------------------------------------
uname -a
Linux spider 2.4.21-4.ELsmp #1 SMP
------------------------------------------------
Both the machines is having more or less same issue..



Comment 3 Rajdeep Sengupta 2005-08-03 10:06:55 UTC
ALso I forgot to attach the autofs versions..

The autofs version for 7.2 client machine is
autofs-3.1.7-21
The autofs version for the EL3 client machine is
autofs-3.1.7-41

The output of /etc/auto.master is
/delsoft yp:auto.home -o soft,intr --debug



Comment 4 Jeff Moyer 2005-08-03 12:31:19 UTC
Your autofs version is very old.  Can you upgrade to the latest RHEL update?

Comment 5 Jeff Moyer 2005-08-03 12:32:17 UTC
One further question:  is the file system mounted when the "file" command fails?
 Does it always fail on the initial access to the mount (i.e. when it triggers
an automount)?

Comment 6 Rajdeep Sengupta 2005-08-03 13:04:31 UTC
First of all, we cannot change the version of redhat, the reason being, we 
send release to the customer on this well defined platforms, and we cannot 
change it.
Answer to 2nd question:
Yes, the file system remains mounted when the "file" command fails, that I 
have checked in the debug file. Also you are correct, it fails on the initial 
access to the mounts even if the filsystem is mounted for a long time. Once it 
gives the i/o error, then the nect time, it gives no error and works fine. 
Then if you continue to access the file, it works fine. But then if you try to 
access another file in the same directory, it gives error for the first time 
again.
ALso if I remove the "soft intr" from my auto.master, then it hangs instead of 
giving i/o error. It hangs till I reboot the machine. Even restarting autofs 
doesnot work.


Comment 7 Jeff Moyer 2005-08-03 13:36:10 UTC
OK, so let's see if I have this all straight.

You did not change anything on the client side.
The NFS servers were upgraded to RHEL 4.
Then you started seeing these problems.

Is that right?  If so, can you trigger the problem without the automounter? 
I.e. mount a share by hand, and see if the first access will give you the error?

I'm putting Steve back on the CC list.  This is really sounding like an NFS
issue, to me.

Comment 8 Rajdeep Sengupta 2005-08-03 14:09:48 UTC
Yes, I didnot change anything in the client side atleast for 7.2 machine. 
Though I tried a higher version of autofs in EL 3 machine, but got the same 
error.
You are correct, as soon as I upgraded my NFS server to rhel 4, all these 
problem started.
Now I have physically mounted it on a temp directory /a and then when I 
use "file" command, it hangs for ever..
--------------------------------------------------
[root@colorado /]# mount elephant:/home6/analyzer /a
[root@colorado /]# file /a/OVL/3.5.0_ovl_2/cheetah/lib/libspyglassVE-static-
SunOS5.a 
------------------------------------------------------
I need to kill the shell..

Comment 9 Jeff Moyer 2005-08-03 14:12:09 UTC
Thanks very much for the clarification and the quick testing turn-around.

Steve, this is most certainly not an autofs issue.  I'm changing the component
to kernel and reassigning to you.

Comment 10 josip 2005-11-23 15:15:55 UTC
Try running NFS clients with iptables disabled.  If the problem goes away, then
this bug may be a duplicate of
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=171267

Bug 171267 seems to be triggered when (after successful authorization) the
client connects to the server to read the filesystem superblock.  The client
sends a TCP SYN to establish a connection, but the server (sometimes) responds
with a pure ACK without the SYN bit; this is blocked by iptables, so the
superblock read fails.  On my systems, this bug is often triggered by client
reboots.


Comment 11 Rajdeep Sengupta 2006-07-03 14:28:33 UTC
We donot enable iptables in our network, so it is disabled by default.
So bug 171267 is not our case atleast.


Comment 12 Horst H. von Brand 2006-12-24 15:43:09 UTC
This happens here to us too. Both client and server are CentOS 4.4 machines.
Client is x86_64 kernel-2.6.9-42.0.3.EL, server is i686, kernel-2.6.9-34.0.2.EL
(yes, just noticed; will update ASAP), everything else is up to date. iptables
is enabled, can't take it down.

Comment 14 Jiri Pallich 2012-06-20 16:02:14 UTC
Thank you for submitting this issue for consideration in Red Hat Enterprise Linux. The release for which you requested us to review is now End of Life. 
Please See https://access.redhat.com/support/policy/updates/errata/

If you would like Red Hat to re-consider your feature request for an active release, please re-open the request via appropriate support channels and provide additional supporting details about the importance of this issue.


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