Reply is bottom-posted. <div class="gmail_quote">On Wed, Mar 18, 2009 at 16:57, Andrew Davis <<a href="mailto:nccomp@gmail.com">nccomp@gmail.com</a>> wrote: <blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"> <div bgcolor="#ffffff" text="#000099"> If I'm reading this correctly, the line about "NRPE daemon cannot be run as user/group root!" is directly from the source code of NRPE. Its not an xinetd thing. I've confirmed that xinetd is running and listening on port 5666. I tried changing the owner/group from nobody:nobody to another unprivileged user, but it didn't work. Same results. It appears that despite my configuring the /etc/nagios/nrpe.cfg and the /etc/xinetd.d/nrpe files to use a user other than root, it still tries to start it as the root user and thus when an incoming connection comes in, it gives the "NRPE daemon cannot be run as user/group root!" error. Any thoughts on how to rectify this? Since NRPE is working fine on Linux, is this just a Mac OS X thing? Any help would be immensely appreciated. AD<div><div></div><div class="h5"> Andrew Davis wrote: <blockquote type="cite"> FYI: /var/log/system.log on the client shows: Mar 18 16:08:07 shu xinetd[29066]: START: nrpe pid=557 from=10.1.1.170 Mar 18 16:08:07 shu nrpe[557]: Error: NRPE daemon cannot be run as user/group root! whether I do the default test (with SSL) or use the -n flag to test w/o SSL. The odd thing is that the nrpe config in /etc/xinetd.d is set to run as nobody:nobody and /etc/nagios/nrpe.cfg is owned by nobody:nobody. Only /usr/local/sbin/nrpe is owned by root (as it should be), but is also set to 755 perms. I've compared to a Linux box I have with NRPE and xinetd working properly and the permissions are identical. I'm stumped... Andrew Davis wrote: <blockquote type="cite">I have two Mac OS X servers, one running 10.3, the other running 10.4. Neither can be upgraded to 10.5 due to third party s/w constraints. Both are PPC based XServe's. Trying to compile nrpe with: <blockquote>./configure --sysconfdir=/etc/nagios --enable-ssl </blockquote> Initially, I got the "cannot find ssl libraries" error: <blockquote>~ checking for SSL headers... SSL headers found in /usr/local/ssl checking for SSL libraries... configure: error: Cannot find ssl libraries </blockquote> I downloaded the latest openssl and built it with: <blockquote>./config --prefix=/usr/local shared --openssldir=/usr/local/openssl make make test make install </blockquote> I then had to edit ~/src/nrpe/configure and change the reference from libssl.so to libssl.dylib After that, nrpe compiled cleanly and I was able to move ~src/nrpe/src/nrpe to /usr/local/sbin and start xinetd up. I've confirmed that port 5666 is open and xinetd is running: <blockquote>/usr/local/src/nrpe-2.12/src root# ps waux|grep xinet|grep -v greproot 29066 0.0 -0.0 27484 308 ?? Ss 3:53PM 0:00.02 /usr/sbin/xinetd -pidfile /var/run/xinetd.pid -stayalive /usr/local/src/nrpe-2.12/src root# netstat -an|grep 5666tcp4 0 0 *.5666 *.* LISTEN </blockquote> However, when connecting from the remote server, I get: <blockquote>/usr/local/nagios/libexec/check_nrpe -H <a href="http://host.mydomain.org" target="_blank">host.mydomain.org</a> CHECK_NRPE: Error - Could not complete SSL handshake. </blockquote> The same test but w/o SSL gives yields: <blockquote>[nagios@nephilim src]$ /usr/local/nagios/libexec/check_nrpe -n -H <a href="http://host.mydomain.org" target="_blank">host.mydomain.org</a> CHECK_NRPE: Received 0 bytes from daemon. Check the remote server logs for error messages. </blockquote> So two questions: 1) I'm a UNIX guy, but obviously Mac's are A) different and B) a tad different being BSD-based. So what's the proper way to stop/restart the xinetd daemon? 2) Any thoughts on SSL handshake error? I've googled it, but I'm not getting very far. Anyone have a step-by-step for compiling nagios plugins and NRPE from source on OS X 10.x (specifically 10.3 and 10.4)? I'm using NRPE for all other internal hosts, so I prefer to use it for the Mac's too. I know I could do it via check_by_ssh and get around this, but I prefer to use NRPE if I can. <pre cols="72">-- </pre></blockquote></blockquote></div></div></div></blockquote><div> On a Mac, your xinetd is a bolt-on over the launchd that's there by default; you've obviously got it running. Since you're in /etc/xinetd.d/<something>, you need to cnfigure a different username via xinetd's config. Look for a /etc/xinetd.d/nrpe file, or similar, containing the config for your nrpe service. I tend to grep for the port number in order to find the file. Remember to check /local/* The time service has an example with juicy comments: service time { # This is for quick on or off of the service disable = yes ... ... # External services must fill out the following # user = # group = ... ... } Take a look there, see if you can choose a better username and/or group and if your port of xinetd honours it. I don't know if you have a nrpe user, or run it as nobody. A better option would be a proper launchd config, allowing you to shutdown xinetd if you're installing it there for this purpose only, but then it's a Mac-only thing, and would be more difficult to maintain for non-Mac people. Allan </div></div>