perfparse-devel

[Perfparse-devel] Perparse Method 4 (pipe) - patch proposal

[Daniele C. <dan...@ti...>]

Hello folks!
First of all, I want to congratulate you on the great job doing...

I'm currently using nagios 1.2 and perfparse 0.105.5 on a test 
environment on a FreeBSD 5.3, monitoring some host and servicies.
I chose the pipe method for perfparse and ran perfparsed as root, but I 
noticed that perfparse create the pipe with prw-r----- permissions; as I 
run nagios as unprivileged user, it wasn't able to write to this pipe... 
I take a look at the source code and noticed that perfparse try to 
create a pipe group writable, but can't do because of some restrictive 
general options set. So I wrote this little patch (in attach) that 
create the pipe with prw-rw---- permissions...
I think was the author intention to create a group writable pipe, 
otherwise don't care about this patch.

I appreciate a lot the summary feature and I'd like to know if it's 
planned a graphical representation and, if yes, if it's planned in a 
short-range or middle-range time.

Daniele

Re: [Perfparse-devel] Perparse Method 4 (pipe) - patch proposal

[Yves <yme...@pe...>]

Hi Daniele,

> Hello folks!
> First of all, I want to congratulate you on the great job doing...

Thanks a lot :)

> I'm currently using nagios 1.2 and perfparse 0.105.5 on a test
> environment on a FreeBSD 5.3, monitoring some host and servicies.

Any compilation problem, or can we consider that it works as perfectly on=
 FreeBSD as on
Linux ?

> I chose the pipe method for perfparse and ran perfparsed as root, but I

First, you should never run a program as root if it does need to. Perfpar=
se does not
need root privileges. This is an admin advice :)

> noticed that perfparse create the pipe with prw-r----- permissions; as =
I
> run nagios as unprivileged user, it wasn't able to write to this pipe..=
.
> I take a look at the source code and noticed that perfparse try to
> create a pipe group writable, but can't do because of some restrictive
> general options set. So I wrote this little patch (in attach) that
> create the pipe with prw-rw---- permissions...
> I think was the author intention to create a group writable pipe,
> otherwise don't care about this patch.

Perfparse creates a pipe with prw-rw---- permissions because I consider t=
hat some users
can run perfparse as "perfparse" user and nagios as "nagios" user, both i=
n the same
group.
It seems to be what you want.

Your problem is a umask problem, and I think this should be set in your e=
nvironment.
Some users like you need a permissive umask like 006. Others prefer a res=
trictive umask
like 066. And I use the usual 022 umask.
There are 2 ways to do it : a config option to specify what permissions t=
o use (a lot of
work) or use "umask" on the command line (nothing to do in perfparse).

I think that most admins will prefer to use umask on the command line bec=
ause it's a
common way to restrict permissions. If you need a specific umask for perf=
parse, this is
easy to do :
#!/bin/sh
umask 006
perfparsed &


Thanks for the patch anyways : it helped at least to understand precisely=
 your problem :)

> I appreciate a lot the summary feature and I'd like to know if it's
> planned a graphical representation and, if yes, if it's planned in a
> short-range or middle-range time.

I let Ben, Tim and others to answer this question.

Yves
--=20
- Homepage    - http://ymettier.free.fr - http://www.logicacmg.com -
- GPG key     - http://ymettier.free.fr/gpg.txt                    -
- Maitretarot - http://www.nongnu.org/maitretarot/                 -
- Perfparse   - http://www.perfparse.org/                          -

Re: [Perfparse-devel] Perparse Method 4 (pipe) - patch proposal

[Ben C. <bcl...@pe...>]

Daniele,

>>I'm currently using nagios 1.2 and perfparse 0.105.5 on a test
>>environment on a FreeBSD 5.3, monitoring some host and servicies.
> 
> Any compilation problem, or can we consider that it works as perfectly on FreeBSD as on
> Linux ?

You may want to go to wiki.perfparse.org and vote for FreeBSD.  These 
polls will give us a good idea of what people use, what they want, and 
where we need to develop in the future :)

>>I appreciate a lot the summary feature and I'd like to know if it's
>>planned a graphical representation and, if yes, if it's planned in a
>>short-range or middle-range time.
> 
> 
> I let Ben, Tim and others to answer this question.

Thanks to Tim for giving us the data.

I plan two graphical representations of this:

1. A pie-chart showing proportion of OK,WARN,CRITICAL,UNKNOWN from the 
total's line.

2. A histergram showing the percent of each of the four states over the 
selected time period.

Yves and my self are discussing methods of rewriting our 'perfchart' 
program to make more generic.  When we have agreed, my self (or somebody 
else?) can have a go at these.

I also plan to use Tim's data to complete an Outage Report.  This will 
display data for a group of Hosts, Services, Groups or All data.  Within 
this collection you will get a summary for the time period, and a 
overall summary.  Again with the same two graphics.

I think that this Outage Report will really give PP a powerful feature.

This is because this is often a model used by IT managers to check 
quality of their network.  The bottom line on quality is the outage 
report.  I note every service I pay for always quotes the '99.99%' 
figure.  Well, now people using PP can prove it!

If we can give that for all services for a whole years data, this is 
something very powerful.

All this summary work also needs to be completed for PostgreSQL, which 
is currently lacking this ability.

Regards,

Ben.