Xymon Mailing List Archive search

HOWTO monitored NFS

8 messages in this thread

list Naudit007 · Mon, 16 Feb 2009 15:49:22 +0100 ·
Hello,

I wish monitored sharing NFS on my servers like classic disks.
What is the best way to monitored NFS and apply the thresholds?
And how to proceed to monitored that ?


When I define the share NFS on "localclient.cfg", it doesn't work.
DISK    /mnt/test   70 95

I have try yet "rpc=nfs" in bb-hosts, but don't have graph and alerts.


Thanks, Regards


-- 

<mailinglist.benoit - at - gmail dot com>
list Stef Coene · Mon, 16 Feb 2009 16:19:15 +0100 ·
quoted from Naudit007
On Monday 16 February 2009, Naudit007 wrote:
Hello,

I wish monitored sharing NFS on my servers like classic disks.
What is the best way to monitored NFS and apply the thresholds?
And how to proceed to monitored that ?


When I define the share NFS on "localclient.cfg", it doesn't work.
DISK    /mnt/test   70 95

I have try yet "rpc=nfs" in bb-hosts, but don't have graph and alerts.
The linux hobbit client uses (file hobbitclient-linux.sh)
df -Pl
And from the manpages:
       -l, --local
              limit listing to local file systems

So remove the "l" and the hobbit client will reporting all file systems.


Stef
list Michael Nemeth · Mon, 16 Feb 2009 10:52:09 -0500 ·
And may hang for a long time if the nfs server is down!

try looking on http://www.deadcat.net/ for a script.
I use one from there with a number of changes to fit my situation .
quoted from Stef Coene

Stef Coene wrote:
On Monday 16 February 2009, Naudit007 wrote:
  
Hello,

I wish monitored sharing NFS on my servers like classic disks.
What is the best way to monitored NFS and apply the thresholds?
And how to proceed to monitored that ?


When I define the share NFS on "localclient.cfg", it doesn't work.
DISK    /mnt/test   70 95

I have try yet "rpc=nfs" in bb-hosts, but don't have graph and alerts.
    
The linux hobbit client uses (file hobbitclient-linux.sh)
df -Pl
And from the manpages:
       -l, --local
              limit listing to local file systems

So remove the "l" and the hobbit client will reporting all file systems.


Stef

list Galen Johnson · Mon, 16 Feb 2009 11:07:41 -0500 ·
I highly recommend against disabling the local only df test.  As Michael points out, it will hang for a long time.  Not only hang but leave behind a LOT of df processes that can only be killed by rebooting the server...at least, kill -9 never works for me.  This is most likely due to the fact that it is waiting on a child to finish.

=G=
quoted from Michael Nemeth

From: michael nemeth [mailto:user-609d3fab5b2d@xymon.invalid]
Sent: Monday, February 16, 2009 10:52 AM
To: user-ae9b8668bcde@xymon.invalid
Subject: Re: [hobbit] HOWTO monitored NFS

And may hang for a long time if the nfs server is down!

try looking on http://www.deadcat.net/ for a script.
I use one from there with a number of changes to fit my situation .

Stef Coene wrote:

On Monday 16 February 2009, Naudit007 wrote:


Hello,


I wish monitored sharing NFS on my servers like classic disks.

What is the best way to monitored NFS and apply the thresholds?

And how to proceed to monitored that ?


When I define the share NFS on "localclient.cfg", it doesn't work.

DISK    /mnt/test   70 95


I have try yet "rpc=nfs" in bb-hosts, but don't have graph and alerts.


The linux hobbit client uses (file hobbitclient-linux.sh)

df -Pl

And from the manpages:

       -l, --local

              limit listing to local file systems


So remove the "l" and the hobbit client will reporting all file systems.


Stef
list Buchan Milne · Mon, 16 Feb 2009 18:14:53 +0200 ·
quoted from Naudit007
On Monday 16 February 2009 16:49:22 Naudit007 wrote:
Hello,

I wish monitored sharing NFS on my servers like classic disks.
Why? Aren't you monitoring the NFS server?

Or, do you really want alerts for 100 servers if one NFS server has low disk 
space?

(maybe you do want it, but IMHO, it usually makes more sense to monitor the 
NFS shares once on the NFS server).
list David Masterson Damaster · Mon, 16 Feb 2009 09:09:59 -0800 ·
Buchan Milne <mailto:user-9b139aff4dec@xymon.invalid> scribbled on Monday,
February 16, 2009 8:15 AM:
quoted from Naudit007
On Monday 16 February 2009 16:49:22 Naudit007 wrote:
I wish monitored sharing NFS on my servers like classic disks.
Why? Aren't you monitoring the NFS server?
What if the NFS server is a NetApp or EMC box?
quoted from Buchan Milne
Or, do you really want alerts for 100 servers if one NFS server has
low disk space?
Not necessarily a lot of servers -- maybe you only detail one server to
keep an eye on the NFS partitions?
quoted from Buchan Milne
(maybe you do want it, but IMHO, it usually makes more sense to
monitor the NFS shares once on the NFS server).
Only if the NFS server allows you to put a monitor on it.  Think *big*
company and turf politics...

David Masterson
list Stef Coene · Mon, 16 Feb 2009 19:36:14 +0100 ·
quoted from Galen Johnson
On Monday 16 February 2009, Galen Johnson wrote:
I highly recommend against disabling the local only df test.  As Michael
points out, it will hang for a long time.  Not only hang but leave behind a
LOT of df processes that can only be killed by rebooting the server...at
least, kill -9 never works for me.  This is most likely due to the fact
that it is waiting on a child to finish.
I only use nfs on AIX and there you can specify on the client that even when 
you do a hard mount, you still can cancel the process that hangs.
I don't know if this exists on linux clients ....


Stef
list Naudit007 · Wed, 18 Feb 2009 13:19:15 +0100 ·
Yes it's just to monitoring a NetApp appliance,  I use currently two servers
for monitore that (with df -P), and work well.
But I search an other best way to monitore my nfs without have an issue type
processes. I can't reboot my hobbit server...

Ok I try it  :

http://www.deadcat.net/search.php?menuosfamily=0&menulicense=0&menusection=0&filtermdate=&filterdescription=nfs&menuNoPerPage=20&action=search&Filter=Filter


2009/2/16 David Masterson (damaster) <user-ad630154120e@xymon.invalid>
quoted from David Masterson Damaster
Buchan Milne <mailto:user-9b139aff4dec@xymon.invalid> scribbled on Monday,
February 16, 2009 8:15 AM:
On Monday 16 February 2009 16:49:22 Naudit007 wrote:
I wish monitored sharing NFS on my servers like classic disks.
Why? Aren't you monitoring the NFS server?
What if the NFS server is a NetApp or EMC box?
Or, do you really want alerts for 100 servers if one NFS server has
low disk space?
Not necessarily a lot of servers -- maybe you only detail one server to
keep an eye on the NFS partitions?
(maybe you do want it, but IMHO, it usually makes more sense to
monitor the NFS shares once on the NFS server).
Only if the NFS server allows you to put a monitor on it.  Think *big*
company and turf politics...

David Masterson

-- 

<mailinglist.benoit - at - gmail!com>