Hello,
I just noticed that on OSX (10.11.1 is what I have now), I can simply
do `brew install rsyslog` and I'm getting rsyslog 7.4.5. Nice. Then I
looked at the formula and I saw it just compiles the thing and it
works. Double-nice.
So then I thought I can just go ahead and compile my own 8.16 with the
modules I want (mmnormalize and omelasticsearch is what I currently
need). And then maybe contribute the formula back to brew.
For now I just gave up and will go back to my trusty Ubuntu VM (pun
not intended - it really is Trusty :p), but I wanted to share in case
it helps and maybe someone else has ideas:
First of all, dependencies. 8.16 has some more:
- liblogging, which has to be downloaded and compiled (there's no brew
formula, at least now). It complained about missing systemd, so I
needed to disable that while configuring
- uuid -> `brew install ossp-uuid` seems to make it stop complaining
- liblognorm. Again, no brew formula, but downloading, configure,
make, make install just works
At this point (hopefully I didn't miss something - though I did `brew
install rsyslog` beforehand which pulled libestr, json-c and
pkgconfig) my `./configure --enable-mmnormalize
--enable-elasticsearch` worked. However, make didn't and this is where
I got stuck.
First it complained about a --whole-archive option that isn't
supported by my `ld`. OK, so I installed binutils which brought GNU
ld. That didn't seem to work, either. Not sure if I couldn't get the
compiler to use GNU ld (though I linked it to /usr/local/ld, which is
before the original /usr/bin/ld. `which ld` gave me the path to GNU
ld) or GNU ld doesn't support --whole-archive on OSX (which is what
searching the web seems to confirm).
Either way, I resorted to simply grep and sed out all the occurrences
of --whole-archive to what seems to be the OSX equivalent of
-all_load. Same with --no-all-archive, which I replaced with
-noall_load. At this point, after brew-installing aclocal and
automake, and tricking the compiler via symlinks to think that my 1.15
versions are also 1.14, it did seem to go past these errors, but it
stumbled into the next one:
duplicate symbol _cnfcurrfn in:
../runtime/.libs/librsyslog.a(librsyslog_la-glbl.o)
../runtime/.libs/librsyslog.a(librsyslog_la-rsconf.o)
duplicate symbol _cnfcurrfn in:
../runtime/.libs/librsyslog.a(librsyslog_la-glbl.o)
../runtime/.libs/librsyslog.a(librsyslog_la-msg.o)
duplicate symbol _cnfcurrfn in:
../runtime/.libs/librsyslog.a(librsyslog_la-glbl.o)
../runtime/.libs/librsyslog.a(librsyslog_la-queue.o)
duplicate symbol _cnfcurrfn in:
../runtime/.libs/librsyslog.a(librsyslog_la-glbl.o)
../runtime/.libs/librsyslog.a(librsyslog_la-action.o)
duplicate symbol _cnfcurrfn in:
../runtime/.libs/librsyslog.a(librsyslog_la-glbl.o)
../runtime/.libs/librsyslog.a(librsyslog_la-template.o)
ld: 5 duplicate symbols for architecture x86_64
clang: error: linker command failed with exit code 1 (use -v to see invocation)
make[2]: *** [lmcry_gcry.la] Error 1
make[1]: *** [all-recursive] Error 1
make: *** [all] Error 2
To which I didn't find any easy solution so I gave up. If anyone has
any ideas or advice, please let me know.
Thanks and best regards,
Radu
--
Performance Monitoring * Log Analytics * Search Analytics
Solr & Elasticsearch Support * http://sematext.com/
_______________________________________________
rsyslog mailing list
http://lists.adiscon.net/mailman/listinfo/rsyslog
http://www.rsyslog.com/professional-services/
What's up with rsyslog? Follow https://twitter.com/rgerhards
NOTE WELL: This is a PUBLIC mailing list, posts are ARCHIVED by a myriad of
sites beyond our control. PLEASE UNSUBSCRIBE and DO NOT POST if you DON'T LIKE
THAT.
