> OK, but it does fail. What does config.log tell you at this point about the > failure reason? Note that the configure test is still broken; it's linking > with > many unnecessary libraries, and it should only link with libdl to check for > the > presence of dlopen. A problem with one of the other libs could result in a > test failure.
Hi, I've sorted this out now. The configure test of pmacct is obviously a bit.. odd. It seems the problem was never that it couldn't find dlopen(). Look at this part of the full config.log I pasted earlier: configure:2874: gcc -o conftest -O2 -Wl,--export-dynamic conftest.c -ldl -L/usr/local/lib -lpcap -lpfring -L/usr/lib -lmysqlclient 1>&5 /usr/local/lib/libpfring.so: undefined reference to `pfring_enable_hw_timestamp' collect2: ld returned 1 exit status The problem lies with the libpfring.so library. I "cheated" the configure script by installing the normal libpcap & libpcap-dev packages (without pfring support) an ran a configure.. which worked fine. I then switched back to the pfring libpcap and ran a configure, which then "found" dlopen in -ldl (but cached from the previous configure). However the make failed on "undefined reference to `pfring_enable_hw_timestamp'" anyway. I went back to my 32-bit test server (where everything works), and noticed that I had checked out a later SVN of PF_RING on the 64-bit box. So there had been an update during the hour between my tests on the 32-bit and the 64-bit squeeze. A bit embarrassing, heh. Needless to say, the newer SVN version doesn't work on 32-bit either. PF_RING isn't stable and the source is only available via SVN (afaik). I'll take this up with the PF_RING devs instead :) Thanks for your time though! Regards, Johan -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/b08f3f3d67451844a7a8a029fcc71e4c1958567...@win01.ad.deltamanagement.se
