Xymon Mailing List Archive search

I can't get any graph on hobbit

10 messages in this thread

list Figaro Nicolas · Thu, 7 Jul 2005 09:41:48 +0200 ·
Hi, 
After recompiling rrdtool 1.2.10 on aix 5.2 and hobbit 4.0.4, I have
infos in my rrd files (rrd fetch shows some entries). 
But I can't get any graph at the bottom of hobbit pages. 
I tried to run the cgi from command line, but I have the following
message : "xgetenv: Cannot find value for variable QUERY_STRING"

How can I set the QUERY_STRING properly and try to generate my graphs ? 
Thanks 
Nicolas FIGARO
list Henrik Størner · Thu, 7 Jul 2005 10:21:39 +0200 ·
quoted from Figaro Nicolas
On Thu, Jul 07, 2005 at 09:41:48AM +0200, FIGARO Nicolas wrote:
Hi, 
After recompiling rrdtool 1.2.10 on aix 5.2 and hobbit 4.0.4, I have
infos in my rrd files (rrd fetch shows some entries). 
But I can't get any graph at the bottom of hobbit pages. 
Did you also recompile Hobbit ? If you are moving from rrdtool 1.0.x
to 1.2.x you must do a "make distclean; ./configure; make; make install"
of Hobbit because the library interface is different in the two rrdtool
versions.
quoted from Figaro Nicolas
I tried to run the cgi from command line, but I have the following
message : "xgetenv: Cannot find value for variable QUERY_STRING"
QUERY_STRING is set by the webserver when running a CGI program. It the
"query" part of the URL, ie. everything after the '?' mark in the URL.


Regards,
Henrik
list Rolf Schrittenlocher · Thu, 07 Jul 2005 10:29:12 +0200 ·
Dear Nicolas,
quoted from Henrik Størner
Hi, 
After recompiling rrdtool 1.2.10 on aix 5.2 and hobbit 4.0.4, I have
infos in my rrd files (rrd fetch shows some entries). 
But I can't get any graph at the bottom of hobbit pages. 
I tried to run the cgi from command line, but I have the following
message : "xgetenv: Cannot find value for variable QUERY_STRING"

How can I set the QUERY_STRING properly and try to generate my graphs ? 
Try (for Solaris, but should work for Linux and AIX as well I hope)
QUERY_STRING="VAR1=VAL1&VAR2=VAL2..."
export QUERY_STRING
then execute the script from command line

or - if it is a perl script - add a line like
$ENV{'QUERY_STRING'} = "VAR1=VAL1&VAR2=VAL2...";
at the top of the script

where VARx are the variables the script needs and VALx are the values of these variables

Rolf


Rolf Schrittenlocher

HRZ/BDV, Senckenberganlage 31, 60054 Frankfurt
Tel. Sammelnr. LBS: (49) 69 - 798 28830
Fax: (XX) XX XXX XXXXX
LBS: user-1e39a1813094@xymon.invalid

Persoenlich: user-6ea8e907e200@xymon.invalid
Tel: (XX) XX - XXX XXXXX
list Figaro Nicolas · Fri, 8 Jul 2005 13:48:44 +0200 ·
Hi, 

I'm back with some troubles( again ?). 

First of all, if you didn't have fun today, I just spent my last day rebuilding my /usr/local/bin dir I trashed due to a stupid "find  -rm" command. (I can send the complete command if you don't know how to spend the rest of your workday). 

I recompiled a lot of stuff, and at last hobbit. 

I tried to run the hobbitcgi from command line, and here is the result : 
export QUERY_STRING="host=hostname&service=la&disp=hostname&graph=hourly&action=view"
exec ~hobbit/server/bin/hobbitgraph.cgi $CGI_HOBBITGRAPH_OPTS
Content-type: image/png
Expires: Fri, 08 Jul 2005 11:43:46 GMT

Content-type: text/html

<html><head><title>Invalid request</title></head>
<body>Garbage ': 08-Jul-2005 13:38:46' after command:
COMMENT:Updated: 08-Jul-2005 13:38:46</body></html>


The  rrd looks good : 
rrdtool fetch ~hobbit/data/rrd/hostname/la.rrd AVERAGE
<snip>
1120821300: 2.0000000000e+01
1120821600: 1.1990000000e+01
1120821900: 9.0000000000e+00
1120822200: 8.0000000000e+00
1120822500: 8.0000000000e+00
1120822800: 6.6666666667e+00
1120823100: 6.6666666667e+00
1120823400: NaNQ

Any idea ? 

NF 
quoted from Henrik Størner

-----Message d'origine-----
De : Henrik Stoerner [mailto:user-ce4a2c883f75@xymon.invalid] 
Envoyé : jeudi 7 juillet 2005 10:22
À : user-ae9b8668bcde@xymon.invalid
Objet : Re: [hobbit] I can't get any graph on hobbit


On Thu, Jul 07, 2005 at 09:41:48AM +0200, FIGARO Nicolas wrote:
Hi,

After recompiling rrdtool 1.2.10 on aix 5.2 and hobbit 4.0.4, I have 
infos in my rrd files (rrd fetch shows some entries).

But I can't get any graph at the bottom of hobbit pages.
Did you also recompile Hobbit ? If you are moving from rrdtool 1.0.x to 1.2.x you must do a "make distclean; ./configure; make; make install" of Hobbit because the library interface is different in the two rrdtool versions.
I tried to run the cgi from command line, but I have the following 
message :
"xgetenv: Cannot find value for variable QUERY_STRING"
QUERY_STRING is set by the webserver when running a CGI program. It the "query" part of the URL, ie. everything after the '?' mark in the URL.


Regards,
Henrik
list Henrik Størner · Fri, 8 Jul 2005 13:56:24 +0200 ·
On Fri, Jul 08, 2005 at 01:48:44PM +0200, FIGARO Nicolas wrote:
<html><head><title>Invalid request</title></head>
<body>Garbage ': 08-Jul-2005 13:38:46' after command:
COMMENT:Updated: 08-Jul-2005 13:38:46</body></html>
The latest (4.0.4) Hobbit version has some changes to hobbitgraph.cfg
which were needed for RRDtool 1.2.x. If you originally installed an 
older version (4.0, 4.0.2, maybe even 4.0.3) then it will not update
your hobbitgraph.cfg because it's a file that you might customize
yourself.

So the easiest solution - if you haven't done any modifications - is 
to just copy the hobbit-4.0.4/hobbitd/etcfiles/hobbitgraph.cfg to your
~hobbit/etc/directory.


Regards,
Henrik
list Figaro Nicolas · Fri, 8 Jul 2005 14:35:25 +0200 ·
I started with hobbit 4.0.4, and I use the rrdtool 1.2.10. 
I can create an rrd graph using this command line: rrdtool graph ~hobbit/server/www/test.png DEF:avg=la.rrd:la:AVERAGE CDEF:la=avg,100,/ AREA:la#00CC00:"CPU Load Average" -u 1.0

I don't know if I can comment some parts of the hobbitgraph.cfg file, or increase a debug level somewhere . (but the garbage message comes from the rrd_graph_helper.c file of rrdtool, perhaps I should send the problem to the rrd mailing list. ). 
quoted from Henrik Størner
NF 
-----Message d'origine-----
De : Henrik Stoerner [mailto:user-ce4a2c883f75@xymon.invalid] Envoyé : vendredi 8 juillet 2005 13:56
À : user-ae9b8668bcde@xymon.invalid
Objet : Re: [hobbit] RE : [hobbit] I can't get any graph on hobbit


On Fri, Jul 08, 2005 at 01:48:44PM +0200, FIGARO Nicolas wrote:
<html><head><title>Invalid request</title></head> <body>Garbage ': 08-Jul-2005 13:38:46' after command:
COMMENT:Updated: 08-Jul-2005 13:38:46</body></html>
The latest (4.0.4) Hobbit version has some changes to hobbitgraph.cfg which were needed for RRDtool 1.2.x. If you originally installed an older version (4.0, 4.0.2, maybe even 4.0.3) then it will not update your hobbitgraph.cfg because it's a file that you might customize yourself.

So the easiest solution - if you haven't done any modifications - is to just copy the hobbit-4.0.4/hobbitd/etcfiles/hobbitgraph.cfg to your ~hobbit/etc/directory.


Regards,
Henrik
list Figaro Nicolas · Fri, 8 Jul 2005 15:21:11 +0200 ·
I just commented out the "COMMENT" part in hobbitd/hobbitgraph.cgi, and it works. (I can see my graphs on hobbit on AIX). 
Here are the lines commented :   +815  /* #ifdef RRDTOOL12 */
  +816  /*      strftime(timestamp, sizeof(timestamp), "COMMENT:Updated\\: %d-%b-%Y %H\\:%M\\:%S", localtime(&now)); */
  +817  /*#else */
  +818  /*      strftime(timestamp, sizeof(timestamp), "COMMENT:Updated: %d-%b-%Y %H:%M:%S", localtime(&now)); */
  +819  /*#endif */
  +820  /*      rrdargs[argi++] = strdup(timestamp); */

But I don't know why the COMMENT generates an error in rrd on AIX. 
Thanks Henrik and Rolf for the replies. 
NF 
quoted from Figaro Nicolas
-----Message d'origine-----
De : FIGARO Nicolas [mailto:user-a732b870264c@xymon.invalid] Envoyé : vendredi 8 juillet 2005 14:35
À : user-ae9b8668bcde@xymon.invalid
Objet : [hobbit] RE : [hobbit] RE : [hobbit] I can't get any graph on hobbit


I started with hobbit 4.0.4, and I use the rrdtool 1.2.10. 
I can create an rrd graph using this command line: rrdtool graph ~hobbit/server/www/test.png DEF:avg=la.rrd:la:AVERAGE CDEF:la=avg,100,/ AREA:la#00CC00:"CPU Load Average" -u 1.0

I don't know if I can comment some parts of the hobbitgraph.cfg file, or increase a debug level somewhere . (but the garbage message comes from the rrd_graph_helper.c file of rrdtool, perhaps I should send the problem to the rrd mailing list. ). 
NF 
-----Message d'origine-----
De : Henrik Stoerner [mailto:user-ce4a2c883f75@xymon.invalid] Envoyé : vendredi 8 juillet 2005 13:56
À : user-ae9b8668bcde@xymon.invalid
Objet : Re: [hobbit] RE : [hobbit] I can't get any graph on hobbit


On Fri, Jul 08, 2005 at 01:48:44PM +0200, FIGARO Nicolas wrote:
<html><head><title>Invalid request</title></head> <body>Garbage ':
08-Jul-2005 13:38:46' after command:
COMMENT:Updated: 08-Jul-2005 13:38:46</body></html>
The latest (4.0.4) Hobbit version has some changes to hobbitgraph.cfg which were needed for RRDtool 1.2.x. If you originally installed an older version (4.0, 4.0.2, maybe even 4.0.3) then it will not update your hobbitgraph.cfg because it's a file that you might customize yourself.

So the easiest solution - if you haven't done any modifications - is to just copy the hobbit-4.0.4/hobbitd/etcfiles/hobbitgraph.cfg to your ~hobbit/etc/directory.


Regards,
Henrik
list Rolf Schrittenlocher · Tue, 26 Jul 2005 14:06:53 +0200 ·
Dear Henrik,

We haven't installed hobbit yet, still using bb, but we want to move as soon as possible.
We are working with virtual hosts. Our services are linked to virtual hostnames, each machine may have a number of virtual hostnames. So one machine should be able to send reports using different hostnames. In bb we had bbaliasname for that purpose which set a limit of 2 to the number of names usable. There we had to hack bbinc, bb-bbexttab, etc. It would be very helpful (if not already done, I didn't get that information in the docu so far) if hobbit wouldn't have such a limit.

kind regards
quoted from Rolf Schrittenlocher

Rolf Schrittenlocher

HRZ/BDV, Senckenberganlage 31, 60054 Frankfurt
Tel. Sammelnr. LBS: (49) 69 - 798 28830
Fax: (XX) XX XXX XXXXX
LBS: user-1e39a1813094@xymon.invalid

Persoenlich: user-6ea8e907e200@xymon.invalid
Tel: (XX) XX - XXX XXXXX
list Henrik Størner · Sun, 31 Jul 2005 14:29:58 +0200 ·
Hi Rolf,
quoted from Rolf Schrittenlocher

On Tue, Jul 26, 2005 at 02:06:53PM +0200, Rolf Schrittenlocher wrote:
We are working with virtual hosts. Our services are linked to virtual hostnames, each machine may have a number of virtual hostnames. So one machine should be able to send reports using different hostnames. In bb we had bbaliasname for that purpose which set a limit of 2 to the number of names usable. There we had to hack bbinc, bb-bbexttab, etc.
So what you're asking for is that each virtual server can have the full
set of client status messages: cpu, disk, memory and so on - but these
should just be copies of the status messages from the real server, not
data obtained by querying the virtual server ?

Or is this related to clustering ? I know some HP-UX clusters have
their clustered applications tied to a virtual IP-address, which then
follows the software package as it moves to different hosts in the
cluster. So maybe you have a custom extension script running and it can run on any node in the cluster, but you always want it reported on a specific host from your bb-hosts file ?

I think Hobbit can do what you want to do without any code changes
being needed, but I have to understand your request first :-)


Regards,
Henrik
list Rolf Schrittenlocher · Mon, 01 Aug 2005 08:49:07 +0200 ·
Hi Henrik,
quoted from Henrik Størner
Hi Rolf,

On Tue, Jul 26, 2005 at 02:06:53PM +0200, Rolf Schrittenlocher wrote:
We are working with virtual hosts. Our services are linked to virtual hostnames, each machine may have a number of virtual hostnames. So one machine should be able to send reports using different hostnames. In bb we had bbaliasname for that purpose which set a limit of 2 to the number of names usable. There we had to hack bbinc, bb-bbexttab, etc.

So what you're asking for is that each virtual server can have the full
set of client status messages: cpu, disk, memory and so on - but these
should just be copies of the status messages from the real server, not
data obtained by querying the virtual server ?
...

That is part of but not totally what we'd like to have. The problem are extension scripts on clients. An example:
One machine may be nameserver using the name xxxdns.uni-frankfurt.de, webserver using the alias xxxweb.uni-frankfurt.de and databaseserver using xxxdb.uni-frankfurt.de at the same time. Each of these aliases has local extension scripts running which send data to (so far) bb-display. In bb we hacked bbinc.sh to get $MACHINE set to the first alias and we could use bbaliasname to enable a second alias. bbexttab is created dynamically so extension scripts for these two aliases will work. There is no way I see so far (except of starting these scripts by crontab instead of bb-client) to enable more than these two aliases to work correctly with extension scripts.

kind regards
Rolf


Mit freundlichen Gruessen
quoted from Rolf Schrittenlocher

Rolf Schrittenlocher

HRZ/BDV, Senckenberganlage 31, 60054 Frankfurt
Tel. Sammelnr. LBS: (49) 69 - 798 28830
Fax: (XX) XX XXX XXXXX
LBS: user-1e39a1813094@xymon.invalid

Persoenlich: user-6ea8e907e200@xymon.invalid
Tel: (XX) XX - XXX XXXXX