FLEXlm Template and script set

bodoe · Joined: 05 Jun 2007 Posts: 6

Hi all

I have a small problem with displaying graphs. I have two license servers, and can only display graphs for one of them. The error message is

"ERROR: openinf 'XXX/file.rrd': No such file or directory"

The query works fine, verbose query shows correct output. Status for the device is "down", all other devices for the other server have status "up". The device that doesn't work is set up exactly like the other devices that do work.

If I runt the query_lmgrd.pl script with "get" data is displayed correctly, the same goes for "index", "test" and "query".

If this thread isn't dead already, perhaps someone can give me a hint? Would be much appriciated.

super-hornet · Cacti User Joined: 27 May 2007 Posts: 144

Maybe u can try to remove/delete all that FlexLM systems u want to monitor and recreate again.

SH

super-hornet · Cacti User Joined: 27 May 2007 Posts: 144

Oh! 2 things about this script you have to aware....
1. Remember to modify the Data Source Template's MAXIMUM value. It is default to LIMIT to 100 license. If your license servers has more than 100 license, it will show u "nan" and it will not get plotted.

2. I not sure about the rest but I change all the data in the Graph Template to use EXACT value instead...because u either check out/in 1 license or not...not point something license.

SH

Taiphun · Joined: 08 Jan 2007 Posts: 21

Hi,
I'm using v0.0.2. I've got everything configured, but when I do 'perl lmgrd-query.pl test' I get nothing.

I can however, run the command from the script manually:

Taiphun · Joined: 08 Jan 2007 Posts: 21

Okay, got that all working.. I wasn't running the lmgrd poller.

Now the rra's aren't being created.

Laffer · Joined: 24 Aug 2007 Posts: 14

Is it possible if Modules have the same name on different Vendor Daemons, that lmgrd does not properly handle them?

Example:

First I configured only one Licence Daemon and Cacti reported around 80 licence modules (apps).
Then I configured a second one, because we run another daemon for a new version of the same application (UGS NX3 and NX5). Now from the first one I have only 24 records as data query result as opposed to 80 before...

Laffer · Joined: 24 Aug 2007 Posts: 14

I fixed it:

Change the lines in lmgrd-poller.pl

# Well, this is stupid.
my %tmpstat = %licstatus;
$tmpstat{$lserver}{$app."_".$1}{total} = $2;
$tmpstat{$lserver}{$app."_".$1}{package} = $licservers{$lserver}{$lport}{$app};
$tmpstat{$lserver}{$app."_".$1}{inuse} = $3;
$tmpstat{$lserver}{$app."_".$1}{port} = $lport;
%licstatus = %tmpstat;
# print Dumper %licstatus;

mpelikan · Joined: 13 Sep 2007 Posts: 2

Hi,

the lmgrd-poller is really good stuff; I installed the lmgrd-poller-0.0.2 version;
Only one thing isn't very nice;

the lmgrd-poller.pl is running and collecting the flexlm stats of my LICENSE1.
The is picked up from cacti, the graphs are available, everything works very nice :

BUT : I STOP the lmgrd-poller.pl ; but the data from my flexlm LICENSE1 are still
in the TIE::SHARELITE memory cache for hours ...

How long do stale data live?
Is it possible, to clean the cache ; maybe once a day ?

Regards,
Martin

Taiphun · Joined: 08 Jan 2007 Posts: 21

I'm revisiting this again, same problems as first time around - my graphs are all NaN's.

super-hornet · Cacti User Joined: 27 May 2007 Posts: 144

The data stay in the memory forever till u reboot.

I write an script where I use "destroy" command to release those memory instead.

I recently found a limitation/bug/or whatever u want to call it.

In the poller, each time when it start to do polling, it actually "lock" the IPC first and when finish polling, it "Unlock" it.

I think this is good if that share memory is going to be write by many applications because when the memory is lock, others cannot read or write onto it. Other application have to wait till the share memory is unlock before it can be read or write. But for this lmgrd-poller, there is only one application/script that is doing the writing....which is the poller. Thus I notice that during the time the poller is doing "Polling", the lmgrd-query (use by the cacti) will have to wait till the poller to complete. In this situation, the runtime increases.

I tried to comment out the lock and unlock on the lmgrd-poller and lmgrd-query, and it works!

SH

super-hornet · Cacti User Joined: 27 May 2007 Posts: 144

mpelikan · Joined: 13 Sep 2007 Posts: 2

Hi super-hornet,

please send me the script to "release" the shared memory;
I want to run it onca a day by cron, so that stale licenses do not live until
the next reboot.

Thanks!

Regards,
Martin

super-hornet · Cacti User Joined: 27 May 2007 Posts: 144

Script to "destroy" the share memory
---------------------------------------------------------------------------
#!/usr/bin/perl

#use strict;
#no strict 'subs';
use lib "/usr/lib/perl5/site_perl/5.6.1";
use Data::Dumper;
use Tie::ShareLite qw( :lock );
### my $log = '/var/log/lmgrd-poller.log';
my $hostinfo_file = "/usr/share/cacti/scripts/hostinfo.inc";

my %licstatus;
# Use below for the default key
$ipc = tie %licstatus, 'Tie::ShareLite', -key => 2030405,
-mode => 0666,
-create => 'no',
-destroy => 'yes'
or die("Could not tie to shared memory: $!");

---------------------------------------------------------------------------

jacobskp · Joined: 09 Aug 2004 Posts: 13

Gabro · Joined: 04 Jun 2008 Posts: 1

Hello

i have problem with flexlm. Everything seems to be ok, but when i look at the graph, it show statistic only for two days ago. The problem is only with flexlm server template.

thx