[hobbit] RE: Disk monitor issues

Peters, Ron rpeters at columbia.com
Mon May 11 21:26:25 CEST 2009


-----Original Message-----
From: Dominique Frise [mailto:dominique.frise at unil.ch] 
Sent: Thursday, April 30, 2009 11:16 PM
To: hobbit at hswn.dk
Subject: Re: [hobbit] RE: Disk monitor issues

Storer, Raymond wrote:
> Ron, if your volume names have spaces or other "special" characters in them you may need to modify the code to account for them. I am no Linux or regex guru, but here is what I did in the 'hobbitclient-linux.sh' file (located in your hobbit client bin directory):
> 
> echo "[df]"
> /usr/sbin/vdf | sed -e '/^[^    ][^     ]*$/{
> N
> s/[     ]*\n[   ]*/ /
> s/SAN VM Storage 2/SAN_VM_Storage_2/g
> s/SAN VM Storage 3/SAN_VM_Storage_3/g
> s/SAN VM Storage 4/SAN_VM_Storage_4/g
> s/Poseidon Local Storage/Poseidon_Local_Storage/g
> s/SAN VM Storage/SAN_VM_Storage/g
> s/SAN Storage 3/SAN_Storage_3/g
> s/SAN Storage 2/SAN_Storage_2/g
> s/SAN Storage 1/SAN_Storage_1/g
> }'
> 
> 
> Ray
> 
> -----Original Message-----
> From: Peters, Ron [mailto:rpeters at columbia.com]
> Sent: Thursday, April 30, 2009 12:32 PM
> To: hobbit at hswn.dk
> Subject: [hobbit] RE: Disk monitor issues
> 
> No answer so far so here's another question then.  How would I debug the server processing the 'client data'.  Really just looking at the 'df' data.  I've enabled both 'verbose' and 'debug' switches when starting the server and haven't found any issues.  I do see the correct entries in the hobbit.chk file.
> 
> Any thoughts on why this data isn't be graphed/displayed on the web page?
> 
> -----Original Message-----
> From: Peters, Ron [mailto:rpeters at columbia.com]
> Sent: Tuesday, April 28, 2009 3:12 PM
> To: hobbit at hswn.dk
> Subject: [hobbit] Disk monitor issues
> 
> I've been making some updates to the monitoring of our vmware ESX servers and wanted to monitor the disk usage of the datastores attached to the servers.  So, I modified the hobbitclient-linux.sh file and changed the [df] section from using df to using the vmware command vdf.
> 
> Now, the 'client data' link for that host on the hobbit/xymon server displays the correct output:
> 
> [df]
> Filesystem            Size  Used Avail Use% Mounted on
> /dev/sda2             4.9G  1.7G  3.0G  36% /
> /dev/sda1              99M   28M   67M  30% /boot
> /dev/sda5             2.0G  364M  1.5G  20% /var/log
> /vmfs/devices         4.8T     0  4.8T   0% /vmfs/devices
> /vmfs/volumes/044c399a-7d915ca5 403G  223G  179G  55% /vmfs/volumes/Templates-NFS
> /vmfs/volumes/491c687d-e3ad8cc1-fd39-00188b4010b4 127G  561M  126G   0% /vmfs/volumes/esxhost.local
> /vmfs/volumes/49b9b542-21c7ce88-f87f-00188b4010b2 491G  354G  137G  72% /vmfs/volumes/DMZ-01-SAN8-128
> /vmfs/volumes/49bebc3c-29931b02-3380-00188b4010b2 491G  353G  138G  71% /vmfs/volumes/DMZ-02-SAN8-97
> /vmfs/volumes/49bfe3b7-3c37dcbc-65a2-00188b4010b2 491G  563M  491G   0% /vmfs/volumes/DMZ-03-SAN8-77
> 
> However, the actual page displaying the filesystem usage and rrd graph only displays local disk:
> 
> Tue Apr 28 14:56:51 PDT 2009 - Filesystems OK
> 
> Filesystem            Size  Used Avail Use% Mounted on
> /dev/sda2             4.9G  1.7G  3.0G  36% /
> /dev/sda1              99M   28M   67M  30% /boot
> /dev/sda5             2.0G  364M  1.5G  20% /var/log
> 
> 
> What's also very odd is that I have roughly a dozen of these types of boxes and some of them report the full output on the web page but most don't.
> 
> Why would the text data be correct, but the web data be different or a sub-set?  All the clients are getting their full data to the server.  I haven't seen any errors on the server logs relating to specific clients.
> 
> Thoughts?
> 
> Thanks again...
> 

Looking at your disk values, the df command you use is probably not
correct. Xymon need values in Kbytes in order to graph them correctly.
Second point is that you need to tell vdf to unsplit its output.

We use this vdf command(through sudo) on our ESX3-servers:

echo "[df]"
    sudo /usr/sbin/vdf -P | egrep -v 
"\/vmfs\/devices|none|tmpfs|shmfs|unknown|iso9660"

Dominique


To unsubscribe from the hobbit list, send an e-mail to
hobbit-unsubscribe at hswn.dk



The client isn't the problem here.  The script is gathering the data correctly on all of my ESX hosts.  The problem is that the server isn't displaying the datastore graphs for all the hosts, only some of them.  This is odd since all of them are running the same script and are reporting the data the same.

Why would the server only process data for some hosts and not others?




More information about the Xymon mailing list