December 2006 - Xymon - Xymon mailman web

[patch] "config" download area is now a setting
by cgoyard＠cvf.fr 04 Dec '06

04 Dec '06

Hi Henrik, here is a small but convenient patch, that helps keep the etc/ directory of the hobbit server tidy. Something handy would be to have a automatic suffix-with-hostname before opening the file, and fallback on the straight requested filename. That is, requesting bb-mysql.cfg from "foo.bar.com" would search for "bb-mysql.cfg-foo.bar.com" and then for "bb-mysql.cfg". I may add this early next week if you didn't already :). Regards, -- Charles Goyard - cgoyard at cvf.fr - (+33) 1 45 38 01 31

4 4

DEFAULT client settings applied even when a different rule matches earlier in hobbit-clients.cfg
by tomg＠mcclatchyinteractive.com 04 Dec '06

04 Dec '06

I'm trying to set up my hobbit-clients.cfg file so that specific hosts have certain settings, and the remaining hosts that do not match pick up the default settings. However, it seems that the default settings are picked up and applied to everything, no matter what. If I understand what I've read in the hobbit-clients.cfg manpage, the first rule match should be applied and the remainder of the config file is ignored for the host. That has not been my experience, though. For example, I have these two entries in my hobbit-clients.cfg: HOST=%radm.*t.mcclatchyinteractive.com UP 30m LOAD 3.0 5.0 DISK * 90 95 MEMPHYS 100 101 MEMSWAP 50 80 MEMACT 90 97 PROC adm 1 DEFAULT # These are the built-in defaults. UP 30m LOAD 3.0 5.0 DISK * 90 95 MEMPHYS 100 101 MEMSWAP 50 80 MEMACT 90 97 PROC default 1 Right now I have a red procs alert for server radm004t.mcclatchyinteractive.com, because the process "default" is not running. I feel like I'm just missing something simple, but can't figure out what it is. Any tips? -- Tom Georgoulias Systems Engineer McClatchy Interactive

2 2

NCV graphing & \t\n in NCV module
by dominic.young＠surfcontrol.com 03 Dec '06

03 Dec '06

Dear Henrik, Big fan of hobbit and thanks to yourself and all others involved in the great work, very much appreciated. I have a question re the NCV module: --- release/hobbit-4.2.0/hobbitd/rrd/do_ncv.c 2006-08-09 22:10:06.000000000 +0200 +++ hobbitd/rrd/do_ncv.c 2006-09-14 22:49:06.259129035 +0200 Version 4.2.0 fixed a problem described in the below thread: http://www.hswn.dk/hobbiton/2006/09/msg00277.html However if the line above the NCV line of a custom test does not end in a tab then new line you still encounter the same problem. Is there any reason why a tab is necessary. All the documentation specifies that to get the NCV graphs working all output must be on a new line and formatted as follows: Sun Nov 20 09:03:44 CET 2006 inode_cache : 330624 dentry_cache : 40891068 Thanks Dominic Young -- No virus found in this outgoing message. Checked by AVG Free Edition. Version: 7.5.430 / Virus Database: 268.15.6/565 - Release Date: 02/12/2006 21:39 This message has been scanned for viruses by SurfControl on-demand services - www.surfcontrol.com

1 0

Custom RRD graphs
by ger＠fwi.ie 03 Dec '06

03 Dec '06

Hi folks, I have created a custom script for capturing CPU utilisation from a device and updated all the relevant files but I cannot get my rrd files to update correctly. An initial rrd file gets created but the data does not get input into it. I have created a rrd-ext sript which is running and is capturing the incomming status data and by looking at the /tmp files I can see that the echo outputs are being generated correctly every 5 mintes. ----> rrd-ext.sh HOSTNAME="$1" TESTNAME="$2" FNAME="$3" if [ "$TESTNAME" = "acpu" ] then echo "`date`" >> /tmp/hobbit.$HOSTNAME.$TESTNAME echo "$FNAME" >> /tmp/hobbit.$HOSTNAME.$TESTNAME # Analyze the message we got # ACPU=`grep "connected" $FNAME | awk '{print $14}'` ACPU=`grep "The 5" $FNAME | cut -d"(" -f2 | cut -d"." -f1` # The RRD dataset definitions echo "DS:acpu:GAUGE:600:0:100" # The filename echo "acpu.rrd" # The data echo "$ACPU" echo "$ACPU" >> /tmp/hobbit.$HOSTNAME.$TESTNAME fi The value I want to graph is simply a cpu % counter, have I got the above values and setting correct for this ? Any help or pointers appreciated. Output that this produces is below showing that it is getting the data correctly from the status input. ---> /tmp/hobbit.$HOSTNAME.$TESTNAME Wed Nov 29 19:54:07 GMT 2006 /usr/lib/hobbit/server/tmp/rrd_msg_13879 14 Wed Nov 29 19:59:05 GMT 2006 /usr/lib/hobbit/server/tmp/rrd_msg_26569 25 Wed Nov 29 20:04:07 GMT 2006 /usr/lib/hobbit/server/tmp/rrd_msg_2788 17 Wed Nov 29 20:09:12 GMT 2006 /usr/lib/hobbit/server/tmp/rrd_msg_11944 9 Ger.

4 9

problems with multiple hobbit servers
by khanrahan＠charter.net 02 Dec '06

02 Dec '06

I am having a problem monitoring ports and procs on multiple hobbit servers. I have a main server (server A) watching 2 dozen hosts. I setup a 2nd server (server B) to watch server A. I want each of them to report on themselves as well as the other server. I ser hobbitclient.cfg to look like this : BBDISP="0.0.0.0" # IP address of the Hobbit server BBDISPLAYS="10.164.138.30|10.164.53.55" # IP of multiple Hobbit servers. They both report ALMOST everything to the other. I set up certain procs to be watched on server B on the localclients.cfg: PROC sshd 1 5 PORT "LOCAL=%([.:]22)$" state=LISTEN min=1 max=5 color=red TRACK=SSH "TEXT=SSH" They all show up on both servers but I setup the same things on server A in the same places, and it shows up in NEITHER! What am I doing wrong? Kevin

1 0

printer monitoring v. static bb-hosts
by hobbit＠anastigmatix.net 02 Dec '06

02 Dec '06

Hi, I started writing a server extension for CUPS monitoring (not just testing that the CUPS /server/ is there, but really monitoring the queue states). The script can easily learn all of the existing queues and their states from the CUPS server, and report status with $BB for each queue using the queue name as hostname and a few different columns (running, accepting, jobcount). So, it feels kind of redundant to have to go and add all of the queue names to bb-hosts (or have them all treated as ghosts with no way to specify what page they appear on). How difficult would it be to support a kind of place-holder entry in bb-hosts, maybe as simple as a group-only tag with a title and no explicit hosts beneath it: group-only running|accepting|jobcount Printers and then some modified form of the 'status' command in bb: $BB $BBDISP "gstatus Printers laser1.running green $(date)" I bet such a capability would be useful for monitoring a lot of things besides printers. Does it seem feasible? Chapman Flack

2 4

[hobbit] SMTP and FTP Services Red
by PNIXON＠ci.somerville.ma.us 02 Dec '06

02 Dec '06

Did you try putting an entry for Hobbit server into their hosts files? I think the delay is from the Server trying to do a reverse lookup on the IP Address. You may also have to change the order in which they resolve DNS.... in /etc/nsswitch.conf (or whatever is appropriate for your distribution) change from hosts: db files nisplus nis dns to hosts: files nisplus nis dns db _____ From: Geoff Hallford [mailto:geoff.hallford at gmail.com] Sent: Wednesday, November 29, 2006 3:46 PM To: hobbit at hswn.dk Subject: [hobbit] SMTP and FTP Services Red Hello, On a few of our servers the SMTP and FTP services are RED but when I telnet to them, they do show the proper 220 status good. They are quite slow to respond though, taking sometimes 10-15 seconds. Is there anyway to get Hobbit to pause and wait for this? If not, how long does it wait before it shows a red status? Thank you very much for your help. Geoff Hallford

5 8

Hobbit-alert depends?
by dan.mcdonald＠austinenergy.com 01 Dec '06

01 Dec '06

Currently there is a depends construct in bb-hosts with a fair number of limitations. I would like to set up some dependencies in the alerts section instead. For example, I monitor ups's for estimated runtime and actual time on battery. I currently alert whenever a UPS goes on battery, and when runtime is less than 30 minutes. I would like to alert on runtime only when we are on battery. -- Daniel J McDonald, CCIE # 2495, CISSP # 78281, CNX Austin Energy http://www.austinenergy.com

1 0

lsof monitor?
by jkwade＠futurefrontiers.com 01 Dec '06

01 Dec '06

Has anyone written or will Hobbit monitor the number of open connections to an application? As an example, using lsof. Sorry if I'm repeating myself. I asked earlier but didn't get an answer. I thought I'd ask again and see if anyone knows a way to monitor it across the clients. I want to check and alert if an application has to many open connections. ie. >1000 Thanks..James

5 5

HTTP error 0
by lebarber＠gmail.com 01 Dec '06

01 Dec '06

What does "HTTP error 0" signify when given as the result of an http test? An empty page? Thanks, Larry Barber

2 1