Xymon Mailing List Archive search

graph problems

6 messages in this thread

list Gianluca Rossi · Wed, 06 Jul 2005 14:52:57 +0200 ·
Hello,
I have just copied my old big brother data files (history and rrd) to a new hobbit 4.0.4 installation, it seems to work well except some problems with rrd file:


- on all linux client the vmstat graph was empty, and the error message
was:
/dati/hobbit/data/rrd/blu/vmstat.rrd from 127.0.0.1: unknown DS name 'cpu_w'
i have deleted the bigbrother vmstat.rrd, and the new one is correct.


- on all AIX, linux and solaris client the netstat is empty. if i delete the rrd file, it is recreated on aix and linux (still empty however), but not for solaris.

the log for linux machine:
larrd-status.log:2005-07-06 14:39:34 RRD error updating /dati/hobbit/data/rrd/blu/netstat.rrd from 127.0.0.1: expected 13 data source readings (got 16) from 1120653574:102891:105479:0:17502925:398003:0:18201:0:U:U:U:U:40401904:26965995:U:453087:...

for solaris machine:

larrd-status.log:2005-07-06 14:42:09 RRD error updating /dati/hobbit/data/rrd/siebdbp/netstat.rrd from 127.0.0.1: expected 13 data sourcereadings (got 16) from 1120653729:4598699:4703299:0:917679:661345:601878:56101:429:3557031050:4281072706:1677389:801734229:2140806543:1738905102:1990:591967:...


- on all windows client, the "memory" graph is always zero (not empty).

how can i debug this situation?

thanx
-- 
Gianluca
list Gianluca Rossi · Thu, 07 Jul 2005 12:28:22 +0200 ·
quoted from Gianluca Rossi
Gianluca Rossi wrote:
- on all linux client the vmstat graph was empty, and the error message
was:
/dati/hobbit/data/rrd/blu/vmstat.rrd from 127.0.0.1: unknown DS name 'cpu_w'
i have deleted the bigbrother vmstat.rrd, and the new one is correct.
OK, i have seen that this is documented .. sorry
- on all windows client, the "memory" graph is always zero (not empty).
It seems that i have found something wrong in larrd/do_la.c : the first error seems a typo, that zeoes the data collected.
the second is that the function atol return a signed long integer, and cannot report memory sizes > 2Gb.

I have written a small patch, using a specific gcc extension: the function atoll and the long long integer (64 bit).

I am the only one having trouble with memory graph on windows machines???

-- 
Gianluca
list Henrik Størner · Thu, 7 Jul 2005 13:39:40 +0200 ·
quoted from Gianluca Rossi
On Thu, Jul 07, 2005 at 12:28:22PM +0200, Gianluca Rossi wrote:
- on all windows client, the "memory" graph is always zero (not empty).
It seems that i have found something wrong in larrd/do_la.c : the first error seems a typo, that zeoes the data collected.
OK, that one looks like a genuine bug. Thanks for the fix.
the second is that the function atol return a signed long integer, and cannot report memory sizes > 2Gb.
Yes, that is a problem - but I'm not sure it really has any importance
here. The do_la.c code only handles memory statistics *if there is no
bb_memory* extension active on this host. So it only receives the memory reports that the BBNT client embeds in the cpu report - and as
far as I know the BBNT client uses an old Win32 API that cannot handle more than 2 GB of memory.
I am the only one having trouble with memory graph on windows machines???
I think most people are using the bb_memory extension script.

I have written a small patch, using a specific gcc extension: the function atoll and the long long integer (64 bit).
Unfortunately this will cause problems for those who do not use gcc to compile Hobbit. I'll see if I can find a more portable solution.


Regards,
Henrik
list Gianluca Rossi · Thu, 07 Jul 2005 15:13:11 +0200 ·
quoted from Henrik Størner
Henrik Stoerner wrote:
the second is that the function atol return a signed long integer, and cannot report memory sizes > 2Gb.

Yes, that is a problem - but I'm not sure it really has any importance
here. The do_la.c code only handles memory statistics *if there is no
bb_memory* extension active on this host. So it only receives the memory reports that the BBNT client embeds in the cpu report - and as
far as I know the BBNT client uses an old Win32 API that cannot handle more than 2 GB of memory.
BBNT correcty report memory up to 4Gb.
I have also a machine with 8Gb (win32 + PAE ..) and BBNT reports olny 4Gb, i have tried bb-memory with the same result.

Just to return on my original post, with bb-memory extension i see the "netstat" reports from windows, but i still can't view the graph, the error is

RRD error updating /dati/hobbit/data/rrd/ctxdc1/netstat.rrd from 127.0.0.1: expected 13 data source readings (got 16)from 1120740867:16053630:11719046:1:579008:5422023:54611:210255:78:U:U:U:U:1361321631:930771102:U:686131:...

the file netstat.rrd is created, but "rrdtool dump netstat.rrd" show only NaN :(
i attach the message as i see it in the hobbit web page, i see nothing stange there ..
thank
-- 
Gianluca
Attachments (1)
list Henrik Størner · Wed, 13 Jul 2005 10:12:22 +0200 ·
quoted from Gianluca Rossi
On Thu, Jul 07, 2005 at 03:13:11PM +0200, Gianluca Rossi wrote:
Henrik Stoerner wrote:

Just to return on my original post, with bb-memory extension i see the "netstat" reports from windows, but i still can't view the graph, the error is

RRD error updating /dati/hobbit/data/rrd/ctxdc1/netstat.rrd from 127.0.0.1: expected 13 data source readings (got 16)from 1120740867:16053630:11719046:1:579008:5422023:54611:210255:78:U:U:U:U:1361321631:930771102:U:686131:...
Somehow you're getting a "netstat.rrd" file created that is in the wrong
format. Hobbit always generates netstat.rrd files with 16 datasets,
never 13. Could you try running this:

rrdtool dump /dati/hobbit/data/rrd/ctxdc1/netstat.rrd |grep "<name>"

This will show the datasets defined in this RRD file. I suspect that it
has been generate by the old netstat-larrd.pl script from LARRD.


Henrik
list Gianluca Rossi · Wed, 13 Jul 2005 12:59:31 +0200 ·
quoted from Henrik Størner
Henrik Stoerner wrote:
On Thu, Jul 07, 2005 at 03:13:11PM +0200, Gianluca Rossi wrote:
Henrik Stoerner wrote:

Just to return on my original post, with bb-memory extension i see the "netstat" reports from windows, but i still can't view the graph, the error is

RRD error updating /dati/hobbit/data/rrd/ctxdc1/netstat.rrd from 127.0.0.1: expected 13 data source readings (got 16)from 1120740867:16053630:11719046:1:579008:5422023:54611:210255:78:U:U:U:U:1361321631:930771102:U:686131:...

Somehow you're getting a "netstat.rrd" file created that is in the wrong
format. Hobbit always generates netstat.rrd files with 16 datasets,
never 13. Could you try running this:

rrdtool dump /dati/hobbit/data/rrd/ctxdc1/netstat.rrd |grep "<name>"
the file has 16 datasets:

hobbit at bb1:~/data/rrd/ctxdc1$ rrdtool dump netstat.rrd | grep '<name>'
                 <name> udpInDatagrams </name>
                 <name> udpOutDatagrams </name>
                 <name> udpInErrors </name>
                 <name> tcpActiveOpens </name>
                 <name> tcpPassiveOpens </name>
                 <name> tcpAttemptFails </name>
                 <name> tcpEstabResets </name>
                 <name> tcpCurrEstab </name>
                 <name> tcpOutDataBytes </name>
                 <name> tcpInInorderBytes </name>
                 <name> tcpInUnorderBytes </name>
                 <name> tcpRetransBytes </name>
                 <name> tcpOutDataPackets </name>
                 <name> tcpInInorderPackets </name>
                 <name> tcpInUnorderPackets </name>
                 <name> tcpRetransPackets </name>
quoted from Henrik Størner

This will show the datasets defined in this RRD file. I suspect that it
has been generate by the old netstat-larrd.pl script from LARRD.
I am sure that this file was generated by hobbit, because i never had a netstat graph for windows machines.

thanx for your help

-- 
Gianluca