Bug 1100037

Also in packstack/plugins/nagios_910.py, the paths should be /usr/lib64/monitoring 'def _copy_script(**kwargs)'.

Note that this affects both RHEL7 and RHEL6 variants.

Changing these causes installation to operate correctly with 'packstack --allinone'.

(In reply to Lon Hohberger from comment #2)
> Note that this affects both RHEL7 and RHEL6 variants.

RHOS5 for both RHEL6 and RHEL7 are now using the monitoring prefix instead of nagios.

So do we need to handle the condition for the older cases (RHOS3/4)? 
I'm assuming we don't unless its gets backported

There are no plans to update to monitoring-plugins in older releases.

Sorry, I meant, it's okay as long as the change doesn't get backported otherwise we have to add a platform test. I suppose there is no reason such cheery pick would happen.

Lon,

While digging, I found disparities between RHOS5 RHEL6 and RHOS5 RHEL7:

Latest corresponding repos are showing:

RHEL6:
monitoring-plugins-ping-1.5-5.el6.x86_64.rpm   
nagios-3.5.1-4.el6.x86_64.rpm                                      
nagios-plugins-nrpe-2.15-3.el6.x86_64.rpm  

versus 

RHEL7:
monitoring-plugins-nrpe-2.15-2.el7.x86_64.rpm   
monitoring-plugins-ping-1.5-5.el7.x86_64.rpm   
nagios-3.5.1-4.el7.x86_64.rpm 

Consolidating the rpm makes more sense (at least for me) than having to handle such cases in packstack.

Thanks,
Gilles

Sorry, it should be nagios-plugins-nrpe.

The change to monitoring-plugins-nrpe was premature.  I did a build revert the change but apparently forgot to put it in the repo.  I fixed this, and will have the repo updated.

The change comes from:

https://www.monitoring-plugins.org/

However, NRPE is Nagios Remote Plugin Executor, so I incorrectly renamed the sub-package.

If you think it's preferred, I can change the path for nagios-plugins-nrpe back to /usr/lib64/nagios/plugins.  It will require some other specfile changes - since there's no nagios-plugins any more.

However, monitoring-plugins-ping and /usr/lib64/monitoring/plugins/ is the correct location for that binary.

My original goal was to change nagios-plugins-nrpe -> monitoring-plugins-nrpe.  So, I can do that, too.

(In reply to Lon Hohberger from comment #13)
> The change comes from:
> 
> https://www.monitoring-plugins.org/
> 

It makes perfect sense in the context of the monitoring plugin project!
Very interesting evolution of the Nagios plugins making it out through the main door. Thank you for the pointer!

> However, NRPE is Nagios Remote Plugin Executor, so I incorrectly renamed the
> sub-package.
> 
> If you think it's preferred, I can change the path for nagios-plugins-nrpe
> back to /usr/lib64/nagios/plugins.  It will require some other specfile
> changes - since there's no nagios-plugins any more.

I think (I might be wrong!) keeping separated namespaces would be more beneficial on the long run. With such approach, NRPE plugin should be living into nagios directory (/usr/lib64/nagios).

Unfortunately that also means nagios-common has to be separated in two, and a monitoring-common RPM to be created to split /usr/lib64/{nagios,monitoring}


(In reply to Lon Hohberger from comment #15)
> My original goal was to change nagios-plugins-nrpe ->
> monitoring-plugins-nrpe.  So, I can do that, too.

NRPE is structurally a Nagios sub packaged, being Nagios specific events broker/agent. And it's not part of the monitoring-plugins.org list and not likely to become part of it. 

Although it doesn't have to be called plugin, for instance, on Fedora, Shinken has shinken-broker. That said it might be too disruptive for now to change it.

Okay - I'll change it back to match fedora.

Note that only nrpe will use /usr/lib64/nagios.

Ping will use /usr/lib64/monitoring

Go ahead and update packstack and I'll test it ASAP.

Actually, it doesn't even look like you deploy check_ping in the nagios config.

So it might only be the top install line that needs to be changed at this point.

Installation passes with the attached patch.

Created attachment 899582 [details]
One Line fix

I also tested that the nagios bits work (e.g. visited nagios pages).  

We don't actually deploy the ping check from packstack, so this change should be sufficient.

Hi Lon,

Ok so the patch is applied only at RPM build time as temporary workaround.

Also, since rhelosp5 for rhel6 and rhel7 are now using monitoring prefix for ping and nagios for nrpe, I've pushed the following patch for packstack.


https://review.openstack.org/#/c/96347/

Martin said there is a step back because of conflicts with EPEL7 using nagios-plugins.

Sam, 

If Fedora is going on the monitoring-plugins road then shouldn't EPEL7 follow to avoid troubles later?

Thanks

(In reply to Gilles Dubreuil from comment #30)
> Martin said there is a step back because of conflicts with EPEL7 using
> nagios-plugins.
> 
> Sam, 
> 
> If Fedora is going on the monitoring-plugins road then shouldn't EPEL7
> follow to avoid troubles later?

Sorry, I'm missing some context here since there are private comments and such. The route Fedora + EPEL 7 are taking is that will both be taking is to have both available, but renaming nagios-plugins to nagios-enterprises-plugins with monitoring-plugins being provided by monitoring-plugins.

> 
> Thanks

(In reply to Sam Kottler from comment #33)
> Sorry, I'm missing some context here since there are private comments and
> such. 

Hi Sam,

The private comments are basically about our internal patching process for RPMs as workarounds.


> The route Fedora + EPEL 7 are taking is that will both be taking is to
> have both available, but renaming nagios-plugins to
> nagios-enterprises-plugins with monitoring-plugins being provided by
> monitoring-plugins.

Could you please clarify?

(In reply to Gilles Dubreuil from comment #34)
> (In reply to Sam Kottler from comment #33)
> > Sorry, I'm missing some context here since there are private comments and
> > such. 
> 
> Hi Sam,
> 
> The private comments are basically about our internal patching process for
> RPMs as workarounds.
> 
> 
> > The route Fedora + EPEL 7 are taking is that will both be taking is to
> > have both available, but renaming nagios-plugins to
> > nagios-enterprises-plugins with monitoring-plugins being provided by
> > monitoring-plugins.
> 
> Could you please clarify?

Sorry, that last bit was a typo:

The route Fedora + EPEL 7 are taking is that both will be available. nagios-plugins will be renamed to nagios-enterprises-plugins and monitoring-plugins will then provide nagios-plugins. Therefore, users will get upgraded toward the path of the software with the same maintainers if they don't actively choose to use the nagios enterprises version.

(In reply to Sam Kottler from comment #35)

Makes sense. Thanks for the follow up.

performed a regression test on nagios using:
# rpm -qa |grep nagios
nagios-plugins-2.0.1-1.2.el6.x86_64
nagios-3.5.1-6.el6.x86_64
nagios-common-3.5.1-6.el6.x86_64
nagios-plugins-ping-2.0.1-1.2.el6.x86_64
nagios-plugins-nrpe-2.15-4.el6.x86_64

on top of:
# rpm -qa |grep packstack
openstack-packstack-2014.1.1-0.34.dev1219.el6ost.noarch

Based on comment 40 setting flag requires_doc_text to "-".

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

http://rhn.redhat.com/errata/RHEA-2014-0950.html