The original question was.
I have been using the automounter and NFS for a few years on our site.
> One of our nfs servers is in Longmont, Colorado and we are in Herndon, VA.
>
> We also use Clearcase, a config managment/version control system. Clearcase
> uses automounter to mount clearcase databases from the Longmont nfs server.
>
> The problem: Anytime we loose our link to Longmont, Many of our nfs clients
> hang completely with error messages: nfs server not responding still
> trying.
>
> Question: Is there a way that we can tell the automounter to forget about
> the lost nfs server and continue with the other things without interruption
> to our work?
Thanks to all who replied. Please send me email directly if you wish to
discuss further about what I am doing to resolve this issue.
Summary
--------
I received several answers. Most of them asked me to do a soft mount of the
remote clearcase VOBs. This will let the user on the desktop continue with
his other work though he may have lost some remote mounts to a particular
server. SUN manuals do not recommend this though since this will not guarentee
data integrity. (Soft mounts will not wait for the server to come back up
later so that the client can finish the write operations)
One person suggested using amd (Instead of automounter, Interesting, I will
try to pursue this and find out more info)
For now, I am soft mounting the remote clearcase VOBS and have not had any
problems so far. I could have used multisite to replicate the remote VOBs but
our users are not ready to work on replicated VOBS since it requires a change
in the process. Our remote site is replicating our VOBS using multisite though.
To do a soft mount of a clearcase VOB using mktag -vob with the -options
argument giving the nfs mount options. See clearcase mktag -vob man pages.
-Ramesh.
--------------------------------------------------------------------------------
I received several answers, Here they are: listed as 1) 2) 3) 4) etc.
1) Maybe you should consider looking at their MultiSite product.
--
Russ Poffenberger DOMAIN: poffen@San-Jose.ate.slb.com
Schlumberger Technologies ATE UUCP: {uunet,decwrl,amdahl}!sjsca4!poffen
1601 Technology Drive CIS: 72401,276
San Jose, Ca. 95110 Voice: (408)437-5254 FAX: (408)437-5246
2) Unfortunately Ramesh with hard NFS mounts the short answer is no, the only way round the problem is to reboot.
You can get around it using soft mounts but run a very real risk of data corruption.
The NFS protocol was designed to work in this way
ie: The client side will wait around until the end of the world
for the NFS server to come back up. This way when it does the client
can flush its buffers to the server and preserve data integrity.
Cheers
Martin
"I used to work for the Sun Hotline but i'm all right now!! :-)"
--------------------------------------------------------------------------------
Martin J.G. Williams Email: Martin.Williams@research.natpower.co.uk
UNIX Support Group Tel: +44 (0)1793 896177
The Electron Building
National Power Plc.
Windmill Hill Business Park, Whitehill Way, Swindon, Wiltshire, SN5 6PB
--------------------------------------------------------------------------------
3)Unfortunately, there's probably nothing you can do. The automounter
is single-threaded in all of it's incarnations to date. If a single
automount attempt cannot be completed, all subsequent attempts will
queue up behind it. No further automounts will be completed until the
first one goes through.
If the link is down for an extended period, there are some workarounds
that will allow you to keep working until it returns, but they require
restarting the daemon and/or rebooting. Not a very elegant solution...
Now, having said all of this, I would certainly like to be proven wrong!
If you happen to here of a reasonably elegant solution, I would very
much like to hear about it.
We also use Clearcase here, and have encountered some interesting problems
with it. Perhaps we could exchange notes?
Hope this helps.
Steve Ehrhardt stevee@sbei.com
SBE Inc. (510)355-7773
San Ramon,CA
"The opinions expressed are those of the author. His employer would disavow
any knowlege of them, presuming they knew that he had any."
4)From bismark@alta.jpl.nasa.gov Fri Jul 21 12:55 EDT 1995
Date: Fri, 21 Jul 95 10:03:35 PDT
From: bismark@alta.jpl.nasa.gov (Bismark Espinoza)
To: rr@netrix.com
Subject: Re: Automount/NFS question.
Cc: bismark@alta.jpl.nasa.gov
How about having additional automount entries to take
if the first one fails?
Or have the automount label to be softmount, background and
have a specific timeout and number of retries?
5)You might look at using amd instead of automounter for your NFS. There's a
decent workup on it in _Unix System Administrators Handbook_, 2ed., by
Nemeth,Synder,Seebass,Hein.
Other option might be to go for soft mounts instead of hard mounts, but it
seriously increases the risk of file corruption for this situation.
Good luck!
Tox
***********************************************
* Tox Gunn .......tox@remarque.berkeley.edu *
* "Your sanity is not my responsibility!" *
***********************************************
6) From: Vinay Somashekar <vinay@css3s0.engr.ccny.cuny.edu>
To: rr@netrix.com
Subject: Re: Automount/NFS question.
Check the Man pages for the "timeo", "retrans" & "soft" parameters, that should
be of help. Type in ' man mount ' and then use the relevant options in the
automounter maps.
-vinay
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:10:30 CDT