[Nagiosplug-devel] Solaris 10: mysql plugin breaks 'make'

Chris Pepper pepper at reppep.com
Tue May 19 13:56:46 CEST 2009


Ton Voon wrote:
 >
 > On 18 May 2009, at 15:15, Chris Pepper wrote:
 >
 >>> If you are using GCC and your MySQL version was compiled using 
Sun's C compiler (or the other way round), then it is likely that the 
configure options will fail as configure will run mysql_config to get 
the appropriate CFLAGS.
 >>
 >>     This assumes MySQL is installed. Our S10 boxen don't have MySQL, 
so I'd amend as follows:
 >>
 >> If you are using GCC to build the plugins, but your MySQL version 
was compiled using Sun's C compiler (or the other way round) -- or if 
you don't have MySQL installed on Solaris at all -- then it is likely 
that the default build procedure will fail, as configure assumes it get 
appropriate CFLAGS by executing mysql_config.
 >>
 >>     But really, configure shouldn't assume MySQL is present at all.
 >
 > Hi Chris,
 >
 > If mysql_config doesn't exist, then it assumes mysql does not exist 
and then doesn't try to compile check_mysql.

     This doesn't work -- on our S10 systems with no MySQL at all, the 
build fails. Perhaps the MySQL configuration needs to be wrapped in a 
check for availability.

 > I think this is a reasonable course of action. The current logic is:
 >   * if you've been told to include mysql, then try to compile check_mysql
 >   * if you've been told not to include mysql, don't try to compile 
check_mysql
 >   * if you have not specified anything, then see if mysql_config is 
available and if it is the compile check_mysql

     This is a good model, but not current behavior.

 > I think the last step could be enhanced to include trying to compile 
a very small mysql program and if that fails, then do not try 
check_mysql. Patches welcome.

     Sorry, I don't have a fix.

Chris

-- 
Chris Pepper:                <http://cbio.mskcc.org/>
                              <http://www.extrapepperoni.com/>





More information about the Devel mailing list