Closed kingdingaling83 closed 7 years ago
Can you check if there is anything in /usr/local/ncpa/var/log/ncpa_listener.log
?
Also, which version are you installing?
Edit: It looks like our current builds are built against 10.11.x which means it probably won't work on that version. I can look for a lower version of OS X and see if we can't build for it.
just tried to install on a 10.10.5 admin@gry-svr004(/var/log):cat /var/log/system.log | grep ncpa Dec 6 17:44:07 gry-svr004 com.apple.xpc.launchd[1] (com.nagios.ncpa.passive): The UserName key is not supported for non-System services. Dec 6 17:44:07 gry-svr004 com.apple.xpc.launchd[1] (com.nagios.ncpa.passive): The GroupName key is not supported for non-System services. Dec 6 17:44:07 gry-svr004 com.apple.xpc.launchd[1] (com.nagios.ncpa.passive[65428]): Service exited with abnormal code: 1 Dec 6 17:44:13 gry-svr004 com.apple.xpc.launchd[1] (com.nagios.ncpa.passive): Service only ran for 6 seconds. Pushing respawn out by 4 seconds. Dec 6 17:44:17 gry-svr004 com.apple.xpc.launchd[1] (com.nagios.ncpa.passive[65433]): Service exited with abnormal code: 1 Dec 6 17:45:22 gry-svr004.hogarthww.prv sudo[65439]: admin : TTY=ttys000 ; PWD=/private/var/log ; USER=root ; COMMAND=/usr/bin/grep -ir ncpa CDIS.custom DiagnosticMessages accountpolicy.log accountpolicy.log.0.gz accountpolicy.log.1.gz accountpolicy.log.2.gz accountpolicy.log.3.gz accountpolicy.log.4.gz accountpolicy.log.5.gz accountpolicy.log.6.gz alf.log amavis.log apache2 asl authd.log authd.log.0.gz awlogs bluetooth.pklg clamav.log com.apple.clouddocs.asl com.apple.revisiond com.apple.xpc.launchd commerce.log coreduetd.log cups daily.out diskspacemonitor.log displaypolicyd.log displaypolicyd.stdout.log emond fax freshclam.log fsck_hfs.log hdiejectd.log hwmond.log hwmond.log.0.bz2 hwmond.log.1.bz2 hwmond.log.2.bz2 hwmond.log.3.bz2 hwmond.log.4.bz2 hwmond.log.5.bz2 install.log install.log.0.bz2 install.log.1.bz2 install.log.2.bz2 install.log.3.bz2 install.log.4.bz2 install.log.5.bz2 kernel-shutdown.log kernel.log.0.bz2 kernel.log.1.bz2 kernel.log.2.bz2 kernel.log.3.bz2 kernel.log.4.bz2
no log for either
cat /usr/local/ncpa/var/log/ncpa_listener.log cat: /usr/local/ncpa/var/log/ncpa_listener.log: No such file or directory
The second error is interesting, considering that they should be in /Library/LaunchDaemons
which I am under the impression runs it as a system service. Any idea why it would consider it a non-System service? I am going to test this here - I only made minor changes since the last OS X build.
thank you so much. we have a wide range of o.s versions here for apple servers and i'd love to use your agent across all of them.
not sure if it's related but xcode isn't installed
happy to provide more logs or run commands for you if need be
I don't think it has to do with xcode, but let me do a bit of testing and I will get back to you - I should be able to find some older OS X versions to test on and we can figure out what needs to be built.
I've got two quick questions:
64bit
i followed the steps on your documentation
cd /tmp
hdiutil attach /tmp/ncpa-
Thanks for the info, I think there are two issues here that I will address individually:
homebrew
they noticed that the UserName and GroupName values for plist
do not work and since we are now dropping user level inside the Python this is no longer necessary. I am building new versions of the 2.0.0-alpha release that should work with 10.11.5+ that I can link you to once build.I can confirm it works perfectly on Mac OS X 10.11.5
also, as far as i can tell the three O.S versions here 10.6.8 / 10.10.5 / 10.11.5 are all 64bit versions of the O.S so it's strange to me that 10.11.5 is the only one that can handle it being 32bit
also, i'm pretty sure they run 32 bit apps as we still use them for Final Cut Server which is exclusively 32bit
Yeah that is very strange, I know that I have had it install on a 10.10.5 version, in fact I just installed it on our Mac VM host earlier today which is running 10.10.5 and it didn't have issues on there.
I created a new DMG for 2.0.0.a and it's available in the NCPA nightly builds directory since I won't be re-building the 1.8.1 installer. You can try using that, for the 10.10.5 install and see if that solves the issue that was given above.
I was only thinking the 10.6.8 had an issue with it being 32-bit due to the error that it was getting since searching for the Job appears to have crashed: Illegal instruction
error shows that it's due to the system version. I'm downloading a 10.6.8 to try on.
i'll test it on 10.10.5 and let you know
It's working on 10.10.5! thank you so much.
just 10.6.8 left
when using the config wizard in nagios xi for some reason i can't see the interfaces to monitor them or the disks. even tho in the npca web GUI i can seem them being monitored. am i missing a step? screen shots below
Nagios Config for 10.10.5 http://postimg.org/image/g1l420aon/
NCPA 10.10.5 disks http://postimg.org/image/wdv5rqp07/
NPCA 10.10.5. network http://postimg.org/image/wdv5rqp07/
Nagios config for a 10.11.5 machine https://postimg.org/image/fagobzhrl/
installing your new nightly build on 10.6.8 i managed to get a crash report
gry-svr001:NCPA-2.0.0.a admin$ cat /Users/admin/Library/Logs/DiagnosticReports/ncpa_listener_2016-12-08-143037_gry-svr001.crash Process: ncpa_listener [76815] Path: /usr/local/ncpa/ncpa_listener Identifier: ncpa_listener Version: ??? (???) Code Type: X86-64 (Native) Parent Process: launchd [875]
Date/Time: 2016-12-08 14:30:37.308 +0000 OS Version: Mac OS X Server 10.6.8 (10K549) Report Version: 6
Exception Type: EXC_BREAKPOINT (SIGTRAP) Exception Codes: 0x0000000000000002, 0x0000000000000000 Crashed Thread: 0
Dyld Error Message: Library not loaded: /Library/Frameworks/Python.framework/Versions/2.7/Python Referenced from: /usr/local/ncpa/ncpa_listener Reason: image not found
Binary Images: 0x7fff5fc00000 - 0x7fff5fc3bdef dyld 132.1 (???) <69130DA3-7CB3-54C8-ABC5-423DECDD2AF7> /usr/lib/dyld
gry-svr001:NCPA-2.0.0.a admin$
Thank you for the crash report! That'll probably help a lot.
As for the wizard, I believe you are using the old version of the wizard (< 1.4.0) in your Nagios XI so the new NCPA 2 instances aren't reporting. It's an easy fix though, just head over to the Manage Config Wizards page in the admin panel and click "Check for Updates" there should be a version 1.4.1 of the NCPA wizard. You can just click install from the NCPA wizard's row. If it doesn't show up for some reason, I can send you the zip file with the updated wizard manually.
great that worked!
i installed python 2.7
trying to run: launchctl start com.nagios.ncpa.listener ..as root shows this in /var/log/system
Dec 8 17:00:47 gry-svr001 sudo[81265]: admin : TTY=ttys001 ; PWD=/Volumes/NCPA-2.0.0.a ; USER=root ; COMMAND=/usr/bin/su
Dec 8 17:00:53 gry-svr001 com.nagios.ncpa.passive[81281]: 2016-12-08 17:00:53,103 81281 INFO started
Dec 8 17:00:56 gry-svr001 com.nagios.ncpa.listener[81287]: Traceback (most recent call last):
Dec 8 17:00:56 gry-svr001 com.nagios.ncpa.listener[81287]: File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/cx_Freeze-4.3.4-py2.7-macosx-10.6-intel.egg/cx_Freeze/initscripts/Console.py", line 27, in
Having looked at some comments about cx_Freeze and some posts about the improper dynamic linking, so instead of using the bundled version of Python it is trying to use the system's version of Python... which doesn't work. I think that the issue is this:
otool -L /usr/local/ncpa/ncpa_*listener
/Library/Frameworks/Python.framework/Versions/2.7/Python ...
...
I think I have this fixed in the new patch that I am applying to the build process but essentially it should be looking for the Python
binary in the executable path... so it should look like this:
@executable_path/Python
I can do this from the Makefile
and see if it works on my test systems quick and get you a new package to try and install on your 10.6.8 system. If you want to try to change this manually you use the command:
install_name_tool -change /Library/Frameworks/Python.framework/Versions/2.7/Python @executable_path/Python /usr/local/ncpa/ncpa_listener
But you need to have xcode installed I believe.
Can you try the latest DMG build from here: https://assets.nagios.com/downloads/ncpa/nightly/
I've re-built it with the changes I noted in the post above. I'm hoping that the 10.6.8 system will be able to find the python binary now.
Dec 8 21:29:07 gry-svr001 com.nagios.ncpa.listener[88524]: Traceback (most recent call last):
Dec 8 21:29:07 gry-svr001 com.nagios.ncpa.listener[88524]: File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/cx_Freeze-4.3.4-py2.7-macosx-10.6-intel.egg/cx_Freeze/initscripts/Console.py", line 27, in
Do you happen to have openssl
installed and /usr/lib/libffi.dylib
on this system?
sh-3.2# which openssl /usr/bin/openssl
sh-3.2# ls /usr/lib/libffi.dylib /usr/lib/libffi.dylib
Still looks like it's trying to use the wrong Python for running the service... I will have to wait for this 10.6 to get set up to test it more on my end.
I'll try again on a server I've not installed previous builds on and let u know
Will leave this open for your comment but I removed the 2.0.0 milestone and bug labels for the support label. The first bug for 10.7+ systems is fixed and I am not 100% certain yet that we will be able to support systems that are below 10.7 ... I will continue to work on this issue but it won't be stopping the release.
is it the python dependency that is the issue? If it's just a case of having the specific python version i can look into forcing that myself no?
could you list what modules etc the programme needs or would it be too many?
Well it shouldn't actually require Python, it's not using the bundled version of Python that comes with NCPA - when you install it, there should be a Python binary called /usr/local/ncpa/Python
that the services should be using - that has a compressed library file and uses all those .so files in that directory. However, on 10.6.8 for some reason it can't find that version of Python... so it's trying to use the one that is built in. You can install python via homebrew or the python.org installer and then install the prereqs like so:
python -m pip install gevent gevent-websocket flask jinja2 requests pyOpenSSL psutil
But I am not sure if that will work with the frozen version of NCPA that you are installing.
Thanks for all your help
i'm going to use NCPA on all servers/os versions apart from 10.6.8.
I'd like to write plugins/scripts that use NCPS to sent a trap depending on the scrips/plugins. for example wrap an rsync so if it fails it sends an NCPS trap. is there documentation on how to do this?
Sounds good, I figured 10.6.x was a bit of a stretch, officially I'm going to label our builds for 10.7+ for the time being.
I'm not sure, can you be a bit more specific? As for NCPS, did you mean SNMP traps? Or can you explain a bit more/link something so I can see.
I'm going to close this since I removed 10.6.8 from the supported list. If you want to get more info on how to set up scripts, etc then you should make a post on the Nagios support forum since the whole support team can help out then too - with more than just NCPA.
seems to install fine.
but when i check the port is listening nothing
here's the error system.log:Dec 6 17:33:44 gry-svr001 com.apple.launchd[1] (com.nagios.ncpa.passive[6785]): Job appears to have crashed: Illegal instruction system.log:Dec 6 17:33:47 gry-svr001 com.apple.launchd[1] (com.nagios.ncpa.listener[6787]): Job appears to have crashed: Illegal instruction