Alerts 🔗 link

Hi,

I've been using BB for several years, and now when i had to build a new
machine i found Hobbit!! Hobbit is much better than BB, but i have a
question though.

I want to send and email-alert to myself if i get a red alert on any of my
monitored systems on any servicers. How can i do this? Is it possbile to use
some kind of wildcard? I tried HOST=* but it didn't work.

Thanks in advance

/ Martin

On Fri, Nov 11, 2005 at 12:27:58PM +0100, Martin Lyberg wrote:

This in the ~hobbit/server/etc/hobbit-alerts.cfg should do it:

    HOST=*
	MAIL user-74d3530ce3ee@xymon.invalid

If it doesn't work, then please check if you can send mail at all
from the Hobbit server. If you run

    echo "mail test" | mail user-74d3530ce3ee@xymon.invalid

and the mail does not arrive, you'll have to check your mail setup
first.


Regards,
Henrik

On 11/11/05, Henrik Stoerner wrote:

Odd, same syntax that didn't work at first. Seems to work now though.

Thank you!

/ Martin

Henrik,

I have som further questions on this topic. I'm not quite sure of the syntax
here. I want to be notified ONLY if the color is red or purple. Can this be
made on a single line like this:

HOST=*
MAIL root REPEAT=1h COLOR=red,purple

Question 2:

I want to send notifications to other people on certain hosts. How can i add
multiple sites on the same line, this doesn't seem to work:

HOST=www.domain1.com|www.domain2.com|www.domain3.com
MAIL user-e2c0c4f8a224@xymon.invalid COLOR=red,purple,yellow REPEAT=1h RECOVERED
MAIL user-6109888833b7@xymon.invalid COLOR=red,purple,yellow REPEAT=1h RECOVERED

Thanks in advance

/ Martin

On Fri, Nov 11, 2005 at 03:49:42PM +0100, Martin Lyberg wrote:

Yes.

You can either list individual hosts, but then you must use commas
to separate hostnames. Like
   HOST=www.domain1.com,www.domain2.com,www.domain3.com
or you can use regular expressions - these begin with a '%' sign, 
followed by the regex. How you'll define that obviously depends on
the hostnames involved, for your example it might be
   HOST=%www.domain[123].com


Henrik

Greetings,

When compiling the newest snapshot on HP-UX 11.0 with gcc, i get the following error:

CC="/usr/local/bin/gcc" CFLAGS="-g -O -D_REENTRANT -DHPUX -I. -I`pwd`/include -DCLIENTONLY=1" OSDEF="-DHPUX" RPATHOPT="" PCREINCDIR="" SSLFLAGS="" SSLINCDIR="" SSLLIBS="" NETLIBS="-lnsl" BBTOPDIR="/home/hobbit" BBLOGDIR="" BBHOSTNAME="" BBHOSTIP="134.48.22.240" BBHOSTOS="" LOCALCLIENT="no" /usr/local/bin/gmake -C lib client
gmake[1]: Entering directory `/home/hobbit/uploaded/snapshot/lib'
/usr/local/bin/gcc -g -O -D_REENTRANT -DHPUX -I. -I/home/hobbit/uploaded/snapshot/include -DCLIENTONLY=1 -I. -I../include    -c -o osdefs.o osdefs.c
In file included from osdefs.c:16:
osdefs.h:25: error: conflicting types for `snprintf'
/usr/local/lib/gcc-lib/hppa2.0n-hp-hpux11.00/3.3.2/include/stdio.h:493: error: previous declaration of `snprintf'
gmake[1]: *** [osdefs.o] Error 1
gmake[1]: Leaving directory `/home/hobbit/uploaded/snapshot/lib'
gmake: *** [lib-client] Error 2


Any ideas?

thanks,
Adam

As a quick fix, (and a fix for an additional similar problem)  i changed the osdefs.h file to be the following (take note, i had to rename the snprintf and the socklen_t variables):

/*----------------------------------------------------------------------------*/
/* Hobbit monitor library.                                                    */
/* Compatibility definitions for various OS's                                 */
/*                                                                            */
/* Copyright (C) 2002-2005 Henrik Storner <user-e180c691dbbd@xymon.invalid>                 */
/*                                                                            */
/* This program is released under the GNU General Public License (GPL),       */
/* version 2. See the file "COPYING" for details.                             */
/*                                                                            */
/*----------------------------------------------------------------------------*/

#ifndef __LIBBBGEN_OSDEFS_H__
#define __LIBBBGEN_OSDEFS_H__

#include "config.h"

#include <sys/types.h>
#include <stdarg.h>

#ifndef HAVE_SOCKLEN_T
typedef int hobbit_socklen_t;
#endif

#ifndef HAVE_SNPRINTF
extern int hobbit_snprintf(char *str, size_t size, const char *format, ...);
#endif

#ifndef HAVE_VSNPRINTF
extern int vsnprintf(char *str, size_t size, const char *format, va_list ap);
#endif

#endif

Adam Scheblein wrote:

On 11/11/05, Henrik Stoerner wrote:

You can either list individual hosts, but then you must use commas

HOST=www.domain1.com <http://www.domain1.com>,www.domain2.com<http://www.domain2.com>;
,www.domain3.com <http://www.domain3.com>;

Thank you, it works like a charm now. I had a blankspace before my hosts ie:

HOST= domain1.com <http://domain1.com>,etc,etc instead of:
HOST=domain1.com <http://domain1.com>,etcetc

That's why it didn't work before.

/ Martin

Greetings,

I have modified the ps in all my clients so that it also sends out per-process cpu usage, and now am trying to use a regular perl expression in the search for processes with a certain usage percentage.  Here is what i have:
%(8\d{2}\.\d{2})|(9[0-4]\.\d{2}) and %(9[5-9]\.\d{2})|(1\d{2}\.\d{2}) as 2 of my searches

These don't work however, because they pull up everything that is 8.xx 8x.xx 9.xx 9x.xx 1.xx 1x.xx and 1xx.xx  I know these espressions work, because i tested them in a perl application...  all i am trying to make it pull out is 8x.xx 9x.xx and 1xx.xx

Any ideas for a fix (either a patch or another way to do this type of a search)??

thanks,
Adam

On Tue, Nov 15, 2005 at 09:08:28AM -0600, Adam Scheblein wrote:

I think you have an error in that first expression - the "8\d{2}" seems
wrong, if you want it to match number 80-89. It should be "8\d" without
the "{2}".

I'm not sure what data you're feeding it, but I tried with this setup in
hobbit-clients.cfg:

HOST=localhost
        PROC %(8\d\.\d{2})|(9[0-4]\.\d{2}) 0 0
	PROC %(9[5-9]\.\d{2})|(1\d{2}\.\d{2}) 0 0

This will give a red if any of the commands in the "ps" listing has the
percentages 80-94 or 95-100. Now let's test it:

$ bbcmd hobbitd_client --test
Hostname (.=end, ?=dump, !=reload) [localhost]: localhost
Test (cpu, mem, disk, proc): proc
ps command string: 82.31 ls
ps command string:
Process %(8\d\.\d{2})|(9[0-4]\.\d{2}) color red: Count=1, min=0, max=0
Process %(9[5-9]\.\d{2})|(1\d{2}\.\d{2}) color green: Count=0, min=0, max=0

Good, the ps commandline string "82.31 ls" is matched as it should be.

Hostname (.=end, ?=dump, !=reload) [localhost]:
Test (cpu, mem, disk, proc): proc
ps command string: 8.29 ls
ps command string:
Process %(8\d\.\d{2})|(9[0-4]\.\d{2}) color green: Count=0, min=0, max=0
Process %(9[5-9]\.\d{2})|(1\d{2}\.\d{2}) color green: Count=0, min=0, max=0

And the string "8.29 ls" is not matched. 

Hostname (.=end, ?=dump, !=reload) [localhost]:
Test (cpu, mem, disk, proc): proc
ps command string: 100.00 muncher
ps command string:
Process %(8\d\.\d{2})|(9[0-4]\.\d{2}) color green: Count=0, min=0, max=0
Process %(9[5-9]\.\d{2})|(1\d{2}\.\d{2}) color red: Count=1, min=0, max=0

"100.00 muncher" is also matched

Hostname (.=end, ?=dump, !=reload) [localhost]:
Test (cpu, mem, disk, proc): proc
ps command string: 1.00 low
ps command string:
Process %(8\d\.\d{2})|(9[0-4]\.\d{2}) color green: Count=0, min=0, max=0
Process %(9[5-9]\.\d{2})|(1\d{2}\.\d{2}) color green: Count=0, min=0, max=0

But "1.00 low" is now matched.


So either your rules are different from what I've used, or your ps
output is unlike that which I used. I think you'll need to post some
more details for us to figure out what the problem is.


Regards,
Henrik

Henrik Stoerner wrote:

OK -- i checked my rules, and realized that i typed them into the e-mail 
wrong -- here is a direct cut/paste
PROC %(8\d\.\d{2})|(9[0-4]\.\d{2}) 0 0 yellow
PROC %(9[5-9]\.\d{2})|(1\d{2}\.\d{2}) 0 0 red

After i pasted this i realized that i was also pulling IP addresses, so 
i changed the rules to this:
PROC %\s+((8\d\.\d{2})|(9[0-4]\.\d{2}))\s+ 0 0 yellow
PROC %\s+((9[5-9]\.\d{2})|(1\d{2}\.\d{2}))\s+ 0 0 red


Which now work, but seems to be acting strange because when it finds a 
bad process it reports this(keep in mind that these are sorted by cpu 
usage and nothing has been edited/cut out):

red %\s+((9[5-9]\.\d{2})|(1\d{2}\.\d{2}))\s+ (found 7, req. at most 0)

adam     18178 18176 99.63 11:19:16 pts/3       02:00 -bash
root        19     0  0.55  Nov  8  ?           05:29 vxfsd

so i am not sure where the "found 7" is coming from, when clearly there 
is only 1...

Thanks,
Adam