Xymon Mailing List Archive search

Too many line on disk graph

17 messages in this thread

list Michael Nemeth · Tue, 18 Oct 2005 09:46:06 -0400 ·
OK the server seam to be no long  spitting the disk graphs.  This is  true for all source of data, bb-clients  local and remote
clients,.  This is with snapshot done on Oct. 16 night.

-- 
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
|     _p_       Mike Nemeth
|  ___| |_____  email(w) user-609d3fab5b2d@xymon.invalid Work: XXX XXX-XXXX          |><___________)          |               Home Page:http://www.geocities.com/mjnemeth/
|               Work Page:http://faraday.motown.lmco.com:3000/~nemethm/ |               Work Page:http://ortsweb/~mnemeth/ |++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
list Pat Vaughan · Wed, 19 Oct 2005 08:47:52 -0400 (EDT) ·
A couple of us have had the same problem.  Stopping the server, clearing
out the Hobbit server tmp subdirectory and restarting it was suggested.

At first it didn't fix anything (and I did a forced reload on the disk web
page), but after a couple of minutes the graphs went back to being split
as normal.
quoted from Michael Nemeth

OK the server seam to be no long  spitting the disk graphs.  This is
true for all source of data, bb-clients  local and remote
clients,.  This is with snapshot done on Oct. 16 night.

--
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
|     _p_       Mike Nemeth
|  ___| |_____  email(w) user-609d3fab5b2d@xymon.invalid Work: XXX XXX-XXXX
|><___________)          |               Home
Page:http://www.geocities.com/mjnemeth/
|               Work Page:http://faraday.motown.lmco.com:3000/~nemethm/
|               Work Page:http://ortsweb/~mnemeth/
|++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

list Henrik Størner · Wed, 19 Oct 2005 14:58:20 +0200 ·
quoted from Pat Vaughan
On Wed, Oct 19, 2005 at 08:47:52AM -0400, Pat Vaughan wrote:
A couple of us have had the same problem.  Stopping the server, clearing
out the Hobbit server tmp subdirectory and restarting it was suggested.

At first it didn't fix anything (and I did a forced reload on the disk web
page), but after a couple of minutes the graphs went back to being split
as normal.
Bizarre. There's absolutely no reason I can think of why a restart of 
Hobbit would change this behaviour.


Henrik
list Michael Nemeth · Wed, 19 Oct 2005 09:21:43 -0400 ·
Still didn't work for me!  I total  wiped out tmp too.
quoted from Henrik Størner


Henrik Stoerner wrote:
On Wed, Oct 19, 2005 at 08:47:52AM -0400, Pat Vaughan wrote:
 
A couple of us have had the same problem.  Stopping the server, clearing
out the Hobbit server tmp subdirectory and restarting it was suggested.

At first it didn't fix anything (and I did a forced reload on the disk web
page), but after a couple of minutes the graphs went back to being split
as normal.
   
Bizarre. There's absolutely no reason I can think of why a restart of 
Hobbit would change this behaviour.


Henrik

-- 
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
|     _p_       Mike Nemeth
|  ___| |_____  email(w) user-609d3fab5b2d@xymon.invalid Work: XXX XXX-XXXX          
|><___________)          |               Home Page:http://www.geocities.com/mjnemeth/
|               Work Page:http://faraday.motown.lmco.com:3000/~nemethm/ 
|               Work Page:http://ortsweb/~mnemeth/ 
|++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
list Michael Nemeth · Wed, 19 Oct 2005 09:36:08 -0400 ·
Oh Henrink I should have added  IS there a way  to  cause the graphs to split up ?  Say to 3 per graph?
Also it this is for the disk and not the trends.


Aslo an  aside ,  I wonder if it would be for  there to be a toggle on the disk graph page to change service=disk to service=disk1
Easy enought to do manually but you  know users.
quoted from Michael Nemeth

Henrik Stoerner wrote:
On Wed, Oct 19, 2005 at 08:47:52AM -0400, Pat Vaughan wrote:
 
A couple of us have had the same problem.  Stopping the server, clearing
out the Hobbit server tmp subdirectory and restarting it was suggested.

At first it didn't fix anything (and I did a forced reload on the disk web
page), but after a couple of minutes the graphs went back to being split
as normal.
   
Bizarre. There's absolutely no reason I can think of why a restart of Hobbit would change this behaviour.


Henrik

-- 
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
|     _p_       Mike Nemeth
|  ___| |_____  email(w) user-609d3fab5b2d@xymon.invalid Work: XXX XXX-XXXX          |><___________)          |               Home Page:http://www.geocities.com/mjnemeth/
|               Work Page:http://faraday.motown.lmco.com:3000/~nemethm/ |               Work Page:http://ortsweb/~mnemeth/ |++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
list Anatoli Bogajewski · Wed, 19 Oct 2005 17:39:50 +0200 ·
Hello there,

is there an option to disable specific tests, esp. disks, from hobbit client?  What i want to do is to replace disk test - since there no option actually to support alerting based on absolute values - by custom script. 
I've tried to modify your source code to support status calculation based on space used, but although in my imagination it was an easy task - i would need to take some c lessons first. :-)

Many thanks,
Anatoli
list Henrik Størner · Wed, 19 Oct 2005 21:53:30 +0200 ·
quoted from Anatoli Bogajewski
On Wed, Oct 19, 2005 at 05:39:50PM +0200, user-4d3800b5a33a@xymon.invalid wrote:
is there an option to disable specific tests, esp. disks, from hobbit client?  What i want to do is to replace disk test - since there no option actually to support alerting based on absolute values - by custom script. 
Not really. You can do it by tweaking the client script so it doesn't
send in a any "df" data, but that's pretty ugly.
I've tried to modify your source code to support status calculation based on space used, but although in my imagination it was an easy task - i would need to take some c lessons first. :-)
It's in the works.


Regards,
Henrik
list Pat Vaughan · Thu, 20 Oct 2005 08:58:07 -0400 (EDT) ·
Again this morning, it's back to one huge graph instead of little ones.

I know someone else asked this, but is there a way to define a graph so
that the display is broken up into smaller graphs (like the way disk
graphing is supposed to work)?  I have some 32 port fibre switches that I
capture error rates on each port and graph them, but the graph ends up
being really big and hard to read.
quoted from Michael Nemeth
A couple of us have had the same problem.  Stopping the server, clearing
out the Hobbit server tmp subdirectory and restarting it was suggested.

At first it didn't fix anything (and I did a forced reload on the disk web
page), but after a couple of minutes the graphs went back to being split
as normal.

OK the server seam to be no long  spitting the disk graphs.  This is
true for all source of data, bb-clients  local and remote
clients,.  This is with snapshot done on Oct. 16 night.

--
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
|     _p_       Mike Nemeth
|  ___| |_____  email(w) user-609d3fab5b2d@xymon.invalid Work: XXX XXX-XXXX
|><___________)          |               Home
Page:http://www.geocities.com/mjnemeth/
|               Work Page:http://faraday.motown.lmco.com:3000/~nemethm/
|               Work Page:http://ortsweb/~mnemeth/
|++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

list Pat Vaughan · Thu, 20 Oct 2005 09:02:02 -0400 (EDT) ·
Another possible clue, when I came in this morning there was a yellow
state on the disk test and Hobbit was displaying one huge disk graph.  I
even did a reload on the page to make sure Hobbit had to regenerate it. 
After I fixed the error and the state went back to green, the graphs were
back to normal.  It might just be a coincidence, though.
quoted from Pat Vaughan
A couple of us have had the same problem.  Stopping the server, clearing
out the Hobbit server tmp subdirectory and restarting it was suggested.

At first it didn't fix anything (and I did a forced reload on the disk web
page), but after a couple of minutes the graphs went back to being split
as normal.

OK the server seam to be no long  spitting the disk graphs.  This is
true for all source of data, bb-clients  local and remote
clients,.  This is with snapshot done on Oct. 16 night.

--
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
|     _p_       Mike Nemeth
|  ___| |_____  email(w) user-609d3fab5b2d@xymon.invalid Work: XXX XXX-XXXX
|><___________)          |               Home
Page:http://www.geocities.com/mjnemeth/
|               Work Page:http://faraday.motown.lmco.com:3000/~nemethm/
|               Work Page:http://ortsweb/~mnemeth/
|++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

list Al Jeffcoat · Thu, 20 Oct 2005 11:43:07 -0400 ·
Hmmm... During the time that the graphs were "conglomorated" for me
also, I too had disk events... Actually, I had a red and a yellow event.
The yellow event was real, and the timing of fixing it coincided with
cleaning the tmp directory, and restarting hobbit.  Strange thing
though, now that I am looking around.  During this yellow disk event,
hobbit also appeared to ignore the disk rules in hobbit-clients.cfg for
another client, which caused a red event for this client.  When the
yellow event was fixed, the red was fixed also, and apparently the
graphs.

Al
quoted from Pat Vaughan

-----Original Message-----
From: Pat Vaughan [mailto:user-37bf2569640f@xymon.invalid] 
Sent: Thursday, October 20, 2005 9:02 AM
To: user-ae9b8668bcde@xymon.invalid
Subject: Re: [hobbit] Too many line on disk graph

Another possible clue, when I came in this morning there was a yellow
state on the disk test and Hobbit was displaying one huge disk graph.  I
even did a reload on the page to make sure Hobbit had to regenerate it. 
After I fixed the error and the state went back to green, the graphs
were
back to normal.  It might just be a coincidence, though.
A couple of us have had the same problem.  Stopping the server,
clearing
out the Hobbit server tmp subdirectory and restarting it was
suggested.

At first it didn't fix anything (and I did a forced reload on the disk
web
page), but after a couple of minutes the graphs went back to being
split
as normal.

OK the server seam to be no long  spitting the disk graphs.  This is
true for all source of data, bb-clients  local and remote
clients,.  This is with snapshot done on Oct. 16 night.

--
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
|     _p_       Mike Nemeth
|  ___| |_____  email(w) user-609d3fab5b2d@xymon.invalid Work: XXX XXX-XXXX
|><___________)          |               Home
Page:http://www.geocities.com/mjnemeth/
|               Work
Page:http://faraday.motown.lmco.com:3000/~nemethm/
|               Work Page:http://ortsweb/~mnemeth/
|++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
list John Glowacki · Wed, 02 Nov 2005 17:50:44 -0500 ·
Was there an update on this? When I have a system with red or yellow events, the disk graph combines all the filesystems into one graph instead of multiple graphs. When I get a green disk event it goes back to multiple graphs. I have a seen this on my Debian and Solaris 4.1.2 installs.

John
quoted from Al Jeffcoat

Jeffcoat, Al wrote:
Hmmm... During the time that the graphs were "conglomorated" for me
also, I too had disk events... Actually, I had a red and a yellow event.
The yellow event was real, and the timing of fixing it coincided with
cleaning the tmp directory, and restarting hobbit.  Strange thing
though, now that I am looking around.  During this yellow disk event,
hobbit also appeared to ignore the disk rules in hobbit-clients.cfg for
another client, which caused a red event for this client.  When the
yellow event was fixed, the red was fixed also, and apparently the
graphs.

Al

-----Original Message-----
From: Pat Vaughan [mailto:user-37bf2569640f@xymon.invalid] Sent: Thursday, October 20, 2005 9:02 AM
To: user-ae9b8668bcde@xymon.invalid
Subject: Re: [hobbit] Too many line on disk graph

Another possible clue, when I came in this morning there was a yellow
state on the disk test and Hobbit was displaying one huge disk graph.  I
even did a reload on the page to make sure Hobbit had to regenerate it. After I fixed the error and the state went back to green, the graphs
were
back to normal.  It might just be a coincidence, though.

A couple of us have had the same problem.  Stopping the server,
clearing
out the Hobbit server tmp subdirectory and restarting it was
suggested.
At first it didn't fix anything (and I did a forced reload on the disk
web
page), but after a couple of minutes the graphs went back to being
split
as normal.

OK the server seam to be no long  spitting the disk graphs.  This is
true for all source of data, bb-clients  local and remote
clients,.  This is with snapshot done on Oct. 16 night.

--
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
|     _p_       Mike Nemeth
|  ___| |_____  email(w) user-609d3fab5b2d@xymon.invalid Work: XXX XXX-XXXX
|><___________)          |               Home
Page:http://www.geocities.com/mjnemeth/
|               Work
Page:http://faraday.motown.lmco.com:3000/~nemethm/
|               Work Page:http://ortsweb/~mnemeth/
|++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
list Henrik Størner · Thu, 3 Nov 2005 07:57:08 +0100 ·
quoted from John Glowacki
On Wed, Nov 02, 2005 at 05:50:44PM -0500, John Glowacki wrote:
Was there an update on this? When I have a system with red or yellow 
events, the disk graph combines all the filesystems into one graph 
instead of multiple graphs. When I get a green disk event it goes back 
to multiple graphs.
I think I've got this one nailed down now. Could you try the attached
patch and let me know if it solves the problem ?

To install, save the attachment to /tmp/hobbit-4.1.2-diskgraph.patch,
then
    cd hobbit-4.1.2
    patch -p0 </tmp/hobbit-4.1.2-diskgraph.patch
    make
then either run "make install" as root, or just copy the
hobbitd/hobbitsvc.cgi program to ~hobbit/server/bin/


Regards,
Henrik

-------------- next part --------------
--- lib/htmllog.c	2005/09/21 08:43:18	1.31
+++ lib/htmllog.c	2005/11/03 06:53:38
@@ -11,7 +11,7 @@
 /*                                                                            */
 /*----------------------------------------------------------------------------*/
 
-static char rcsid[] = "$Id: htmllog.c,v 1.31 2005/09/21 08:43:18 henrik Exp $";
+static char rcsid[] = "$Id: htmllog.c,v 1.32 2005/11/03 06:53:20 henrik Exp $";
 
 #include <ctype.h>
 #include <stdlib.h>
@@ -93,6 +93,7 @@
 		if (p) {
 			*p = '\0';
 			fprintf(output, "%s", restofmsg);
+			*p = '&';
 
 			color = parse_color(p+1);
 			if (color == -1) {
list John Glowacki · Thu, 03 Nov 2005 13:40:27 -0500 ·
Thanks Henrik,

This is close. I have 14 filesystems. The first 3 graphs show 4 filesystems each. The forth graph shows 2 filesystems. And there is a 5th graph which is empty.

If I look under trends I see the 4 graphs I would expect to see. The graphs display exactly like the disk web page except I don't see the 5th empty graph.

-John
quoted from Henrik Størner

Henrik Stoerner wrote:
On Wed, Nov 02, 2005 at 05:50:44PM -0500, John Glowacki wrote:
Was there an update on this? When I have a system with red or yellow events, the disk graph combines all the filesystems into one graph instead of multiple graphs. When I get a green disk event it goes back to multiple graphs.

I think I've got this one nailed down now. Could you try the attached
patch and let me know if it solves the problem ?

To install, save the attachment to /tmp/hobbit-4.1.2-diskgraph.patch,
then
    cd hobbit-4.1.2
    patch -p0 </tmp/hobbit-4.1.2-diskgraph.patch
    make
then either run "make install" as root, or just copy the
hobbitd/hobbitsvc.cgi program to ~hobbit/server/bin/


Regards,
Henrik


--- lib/htmllog.c	2005/09/21 08:43:18	1.31
+++ lib/htmllog.c	2005/11/03 06:53:38
@@ -11,7 +11,7 @@
 /*                                                                            */
 /*----------------------------------------------------------------------------*/
 -static char rcsid[] = "$Id: htmllog.c,v 1.31 2005/09/21 08:43:18 henrik Exp $";
+static char rcsid[] = "$Id: htmllog.c,v 1.32 2005/11/03 06:53:20 henrik Exp $";
  #include <ctype.h>
 #include <stdlib.h>
@@ -93,6 +93,7 @@
 		if (p) {
 			*p = '\0';
 			fprintf(output, "%s", restofmsg);
+			*p = '&';
  			color = parse_color(p+1);
 			if (color == -1) {

list Henrik Størner · Thu, 3 Nov 2005 23:49:17 +0100 ·
quoted from John Glowacki
On Thu, Nov 03, 2005 at 01:40:27PM -0500, John Glowacki wrote:
This is close. I have 14 filesystems. The first 3 graphs show 4 
filesystems each. The forth graph shows 2 filesystems. And there is a 
5th graph which is empty.

If I look under trends I see the 4 graphs I would expect to see. The 
graphs display exactly like the disk web page except I don't see the 5th 
empty graph.
The difference in behaviour comes from how the number of disk graphs
is determined; the "disk" view counts the number of lines in the
current status-message, whereas the "trends" view actually counts
the number of disk*.rrd files stored (there is one such file for each
graph).

I think the lines with the red/yellow "warning lights" confuse the
line-counting algorithm used by the "disk" status view. If you do a
"view source" of the disk status page, there is a line like this:

   <!-- linecount=9 -->
   <table summary="disk Graph"><tr><td><A 

near the bottom. The linecount should match the number of filesystems
in your status message, +/- 1.


Henrik
list John Glowacki · Thu, 03 Nov 2005 18:24:43 -0500 ·
quoted from Henrik Størner

Henrik Stoerner wrote:
On Thu, Nov 03, 2005 at 01:40:27PM -0500, John Glowacki wrote:

This is close. I have 14 filesystems. The first 3 graphs show 4 filesystems each. The forth graph shows 2 filesystems. And there is a 5th graph which is empty.

If I look under trends I see the 4 graphs I would expect to see. The graphs display exactly like the disk web page except I don't see the 5th empty graph.

The difference in behaviour comes from how the number of disk graphs
is determined; the "disk" view counts the number of lines in the
current status-message, whereas the "trends" view actually counts
the number of disk*.rrd files stored (there is one such file for each
graph).

I think the lines with the red/yellow "warning lights" confuse the
line-counting algorithm used by the "disk" status view. If you do a
"view source" of the disk status page, there is a line like this:

   <!-- linecount=9 -->
   <table summary="disk Graph"><tr><td><A 
near the bottom. The linecount should match the number of filesystems
in your status message, +/- 1.


Henrik

It has <!-- linecount=18 -->. I have 3 lines with the red/yellow "warning lights", 1 heading line from the df, and the 14 filesystems.

John
list Henrik Størner · Fri, 4 Nov 2005 12:29:54 +0100 ·
quoted from John Glowacki
On Thu, Nov 03, 2005 at 06:24:43PM -0500, John Glowacki wrote:

Henrik Stoerner wrote:
I think the lines with the red/yellow "warning lights" confuse the
line-counting algorithm used by the "disk" status view. If you do a
"view source" of the disk status page, there is a line like this:

  <!-- linecount=9 -->
  <table summary="disk Graph"><tr><td><A 

near the bottom. The linecount should match the number of filesystems
in your status message, +/- 1.
It has <!-- linecount=18 -->. I have 3 lines with the red/yellow 
"warning lights", 1 heading line from the df, and the 14 filesystems.
OK, so it is as I expected. This is fixed in the current snapshot,
available from http://www.hswn.dk/beta/ - or just grab this file
http://www.hswn.dk/beta/snapshot/lib/htmllog.c, copy it on top of
your hobbit-4.1.2/lib/htmllog.c and run "make" + "make install".


Regards,
Henrik
list John Glowacki · Fri, 04 Nov 2005 07:26:05 -0500 ·
quoted from Henrik Størner
OK, so it is as I expected. This is fixed in the current snapshot,
available from http://www.hswn.dk/beta/ - or just grab this file
http://www.hswn.dk/beta/snapshot/lib/htmllog.c, copy it on top of
your hobbit-4.1.2/lib/htmllog.c and run "make" + "make install".


Regards,
Henrik
Henrik,

It looks good. I am getting <!-- linecount=14 --> now.

Thanks,
John