I have a machine monitored by nagios that has the following line in nrpe.cfg
include_dir=/etc/nagios/cfgs/
The 'cfgs' dir consists of following files with various commands.
servers.cfg
database.cfg
regular.cfg
However, some commands are repeated among these cfg files but they differ in the arguments. Eg: check_disk command on database has different threshold and on servers have different threshold for the same command.
My question is: If there are multiple definitions of a command, which command is picked up by nrpe? Is there some way to identify it?
You generally can not have duplication in command names in your nrpe.cfg files - though no error is thrown if there are duplicates. I've had this happen before (accidentally) with very large configs, and nrpd has always chosen to use the very LAST matching command it found. In my case, being a single config file, it was always easy to spot the LAST matching command in the file. With many included files, this could be an issue. I believe it would load these files in alphabetical order with the include_dir command you provided.
In any case, I'd look for more unique names for these duplicates. Such as 'check_disk_db' for the one checking database disks. These names can be just about anything you want them to be - but remember to change them on the Nagios side as well. Using unique command names would make your life as a Nagios admin much easier.
As far as figuring out which is actually being used. You can turn on debugging. The debugging output doesn't include the file the command came from, but does show you what options were passed and exactly which actual commands where executed via the check command. This info should be enough to see which nrpe command was selected.
Related
My current scenario is like this:
I need to login to sqlplus from a shell script to call a stored procedure.
After that I need to create a CSV file by SPOOLING data from a table.
Then I need to check whether the CSV file has been created in a particular directory and depending on the result an update query needs to be run.
I know that this can be checked within sqlplus with the help of UTL_FILE package but unfortunately due to Client policies,the access of this package is restricted in the current system.
Another way is to exit from sqlplus and perform the file check in UNIX and then again log in to sqlplus to perform the rest actions. But this I believe would result in slower execution time and performance is an important factor in this implementation as the tables contain huge volumes of data(in millions).
So is there any other way to check this from sqlplus without exiting from the current session?
System Info:
OS - Red Hat Enterprise Linux
Database - Oracle 11g
If the file is on the same machine that you're running SQL*Plus on, you could potentially use the host command.
If the file you're checking is the same one you're spooling to, it must exist anyway, or you would have got an SP error of some kind; but if you do want to check the same file for some reason, and assuming you have a substitution variable with the file name:
define csv_file=/path/to/spool.csv
-- call procedure
spool &csv_file
-- do query
spool off
host ls &csv_file
update your_table
set foo=bar
where &_rc = 0;
If the file exists when the host command is run, the _rc substitution variable will be set to zero. If the file doesn't exist or isn't readable for any reason it will be something else - e.g. 2 if the file just doesn't exist. Adding the check &_rc = 0 to your update will mean no rows are updated if there was an error. (You can of course still have whatever other conditions you need for the update).
You could suppress the display of the file name by adding 1>/dev/null to the host command string; and could also suppress any error messages by also adding 2>/dev/null, though you might want to see those.
The documentation warns against using &_rc as it isn't portable; but it works on RHEL so as long as you don't need your script to be portable to other operating systems this may be good enough for you. What you can't do, though, is do anything with the contents of the file, or interpret anything about it. All you have available is the return code from the command you run. If you need anything more sophisticated you could call a script that generates specific return codes, but that's getting a bit messy.
Our installer updates our application from using java7 to java8. In doing so, we'd like to update their Tomcat7 configuration to limit the Metaspace at the same level as their PermGen space (which they might have customized). So if they had...
-XX:MaxPermSize=256m
we'd like to set...
-XX:MaxMetaspaceSize=256m
The trick is, the installer must do this via a batch file. (Perhaps querying Tomcat with some command that would spit out the present value to a file, and then read that value in, and then fire off another command to set the value.)
If I can get the present MaxPermSize, I know that last piece of the puzzle:
"%TOMCAT7_EXE%" %SERVICE_NAME% --JvmOptions "-XX:MaxMetaspaceSize=%MAX_SIZE%"
(It might also be nice to remove the no longer applicable MaxPermSize, but that would be just for extra-credit.)
I am trying to use the tool here business objects query builder output
And there are virtually no examples, so I'm struggling to make it work. It produces no errors, but outputs no file in the directory where the batch file is, that I can see.
Here is the code inside querybuilder.bat:
set lib=c:\Program Files\Business Objects\Common\4.0\java\lib
java -cp "querybuilder.jar;poi-3.8-20120326.jar;%lib%\*" org.bukhantsov.querybuilder.Program %*
Here is the code inside what I am running, which I've named RunQuery_ALLACTIVE.bat, except of course with my Server, Username, and Password changed for the purpose of this post.
I have this all on one line, with no line breaks.
querybuilder.bat -cms:SERVER -username:OURUSERNAME -password:OURPASSWORD -query:"SELECT * FROM CI_INFOOBJECTS where SI_SCHEDULE_STATUS = 9 order by SI_NAME" -auth:windowsad -excel "Output.xls"
Can't tell if the - options go on different LINES ?
Can't tell if I'm supposed to put output file in quotes, or if it should be an existing file or not?
can't tell if for Windows AD (which we use), I would put "Windows AD" or WindowsAD, I'm assuming no spaces obviously.
Tons of unanswered questions on this tool - it LOOKS cool, but has anyone actually successfully used it? Can't really find comments or history on the 'net..
To answer your questions:
The options go on the same line, not on different ones
As Joe said, you'll need to specify the output file as -excel:"Output.xls"
If you want to use Windows AD, you'll probably need to specify secWinAD (case-sensitive).
If you're not sure about the command line options, I suggest you build up gradually: first only specify the required options, then add the optional ones one by one so you know which one is giving you problems.
Also, I noticed that the download page contains a version compiled for XI3.x and BI4. Make sure you use the correct version, corresponding to the version of BusinessObjects you're using. Also, verify the path in the batch file to see if it points to a valid folder containing the JAR files for the BusinessObjects environment.
Update:
I just noticed that the same author/developer created another application (GUI, not command line) that might be a bit easier to use. Have a look here.
Description:
I am searching a very large server for files that is on a different server. right now I open command prompt and type
DIR [FILE NAME] /S/4
This returns the server location of the file with some other stuff that is not really needed.
Question:
I have a lot of files to search and one by one input into the above command could take forever. Is there a way I could input all of the names of all the files and only search once and the search results would only need to show file name and location?
First, I hope you don't mean DOS, but rather Windows cmd or batch.
You can certainly write a script that will run your DIR command once per file being sought.
But what you most likely want instead is to search once and print the path of each file found. For this you can use PowerShell's FindChildItem or the improved one posted here: http://windows-powershell-scripts.blogspot.in/2009/08/unix-linux-find-equivalent-in.html
It will be something like:
Find-ChildItem -Name "firstfile.txt|secondfile.txt|..."
Another approach is to install msys or cygwin or another Linux tools environment for Windows and use the Linux find command.
I'm trying to create a GNU like configure script for Windows in a batch file. It seems to work ok so far except depending on the order options get ignored and in particular with the '--with-smtube' option it sometimes gets the path but again depending on the order causes it to result in:
test.cmd --enable-portable --with-smtube=C:\svn\smtube
configure: error: unrecognized option: `C:\svn\smtube'
Try `--with-smtube --help' for more information
(supposed to also say Try `test.cmd --help)
I copied the structure from another script I found but not having success. The order of arguments shouldn't matter and arguments can be omitted (all or some). Can someone steer me in the right direction?
The script is for a Qt program, it uses command prompt as the shell and not msys or cygwin or anything like that.
This is what I have so far: http://redxii.users.sourceforge.net/test.cmd