-feature: Do not sort by snmp port if you only have snmp on one port
-feature: Order Devices by their number of poller items and Device Threads
-issue: Remove legacy setting device_threads_exist
On FreeBSD when configured with default settings, compilation of spine
will always fail:
```
/bin/sh ./libtool --tag=CC --mode=link cc -I/usr/local/include/net-snmp -I/usr/local/include/net-snmp/.. -I/usr/local/include/mysql -g -O2 -L/usr/local/lib -L/usr/local/lib/mysql -o spine sql.o spine.o util.o snmp.o locks.o poller.o nft_popen.o php.o ping.o keywords.o error.o -lnetsnmp -lmysqlclient_r -lm -ldl -lmysqlclient -lm -ldl -lcrypto -lz -lpthread -ldl -lm -lpthread -lssl
libtool: link: cc -I/usr/local/include/net-snmp -I/usr/local/include/net-snmp/.. -I/usr/local/include/mysql -g -O2 -o spine sql.o spine.o util.o snmp.o locks.o poller.o nft_popen.o php.o ping.o keywords.o error.o -L/usr/local/lib -L/usr/local/lib/mysql -lnetsnmp -lmysqlclient_r -lmysqlclient -lcrypto -lz -ldl -lm -lpthread -lssl
ld: error: undefined symbol: backtrace
>>> referenced by error.c:85
>>> error.o:(spine_signal_handler)
ld: error: undefined symbol: backtrace_symbols
>>> referenced by error.c:86
>>> error.o:(spine_signal_handler)
cc: error: linker command failed with exit code 1 (use -v to see invocation)
*** Error code 1
Stop.
```
When the configure script is run it correctly shows that backtraces are
not supported on this platform:
```
...
checking whether we are using traditional popen... no
checking whether to verify net-snmp library vs header versions... no
checking if we can support backtracing... no
checking if we can support mysql/mariadb retry count... no
checking if we can support mysql/mariadb ssl keys... yes
...
```
And yet in config/config.h we have HAVE_EXECINFO_H defined:
```
/* Do we have backtracing capabilities? */
#define HAVE_EXECINFO_H 1
```
This is because of an error in configure.ac where the value of
HAVE_EXECINFO_H is set twice, the first time by
AC_CHECK_HEADERS(execinfo.h) which returns true (The FreeBSD system does
have execinfo.h, it just does not support the function used here) and a
2nd time later on by a AC_LINK_IFELSE:
```
# See if we can support backtracing
AC_MSG_CHECKING([if we can support backtracing])
AC_LINK_IFELSE([AC_LANG_PROGRAM([[
#include <stdlib.h>
#include <execinfo.h>
]], [[
void *array[10];
size_t size;
// get void*'s for all entries on the stack
size = backtrace(array, 10);
if (size) {
exit(0);
} else {
exit(1);
}
]])],[ AC_MSG_RESULT(yes)
AC_DEFINE(HAVE_EXECINFO_H,1,[Do we have backtracing capabilities?])
],[AC_MSG_RESULT(no)
])
```
Even though the above AC_LANG_PROGRAM fails on FreeBSD, because of the
earlier AC_CHECK_HEADERS succeeding HAVE_EXECINFO_H is already defined
and remains that way leading to the compilation failure.
Since the AC_LANG_PROGRAM check here seems more thorough then the simple
header check and will fail anyway if execinfo.h does not exist I think
the correct course of action here is to remove the AC_CHECK_HEADERS
macro for this particular header completely.
Co-authored-by: Erik Sejr <eriks@ssimicro.com>
Probably a copy/paste error, it does not impact the functionality of the
script, just throws an error when it is run.
Co-authored-by: Erik Sejr <eriks@ssimicro.com>
netniV