Created attachment 405088 [details, diff] In file included from sql.c:34:0: common.h:140:39: fatal error: ucd-snmp/ucd-snmp-config.h: No such file or directory As in summary. It cannot compile because cacti-spine 0.8.8d is broken in some areas (I attach patches for resolve all problems). After all that fixes cacti-spine start to compile and works like a charm later.
Created attachment 405090 [details, diff] ld: locks.o: undefined reference to symbol 'pthread_once@@GLIBC_2.2.5'
Created attachment 405092 [details, diff] SPINE: Poller[0] FATAL: MySQL Error:'1054', Message:'Unknown column 'polling_time' in 'field list'' (Spine thread)
Created attachment 405094 [details, diff] adds patches to 0.8.8d ebuild
Created attachment 405096 [details, diff] adds patches to 0.8.8d ebuild
(In reply to Tomasz Chilinski from comment #2) > Created attachment 405092 [details, diff] [details, diff] > reverts unneeded commit made by upstream You seem to know something we don't?
Comment on attachment 405092 [details, diff] SPINE: Poller[0] FATAL: MySQL Error:'1054', Message:'Unknown column 'polling_time' in 'field list'' (Spine thread) I guess you meant to fix this: SPINE: Poller[0] FATAL: MySQL Error:'1054', Message:'Unknown column 'polling_time' in 'field list'' (Spine thread)
(In reply to Jeroen Roovers from comment #5) > (In reply to Tomasz Chilinski from comment #2) > > Created attachment 405092 [details, diff] [details, diff] [details, diff] > > reverts unneeded commit made by upstream > > You seem to know something we don't? Upstream actually has made commit: http://svn.cacti.net/viewvc/spine/tags/0.8.8d/poller.c?r1=7495&r2=7679 (this is my attachment content) which shouldn't go to 0.8.8d branch, because it uses database fields which are not present in cacti-0.8.8d. If we want to have spine 0.8.8d usable, we should revert it and then spine compiles flawlessly. We can wait for 0.8.8e (I suppose they'll release it soon).
Comment on attachment 405092 [details, diff] SPINE: Poller[0] FATAL: MySQL Error:'1054', Message:'Unknown column 'polling_time' in 'field list'' (Spine thread) PATCH COMMAND: patch -p1 -g0 -E --no-backup-if-mismatch < '/newaches/gentoo/cvs/gentoo-x86/net-analyzer/cacti-spine/files/cacti-spine-0.8.8d-polling_time.patch' ================================================= checking file poller.c Hunk #1 FAILED at 168. Hunk #2 FAILED at 217. Hunk #3 FAILED at 287. Hunk #4 FAILED at 376. Hunk #5 succeeded at 467 with fuzz 2 (offset 16 lines). Hunk #6 FAILED at 501. Hunk #7 FAILED at 516. Hunk #8 FAILED at 530. Hunk #9 succeeded at 1296 with fuzz 2 (offset 34 lines). Hunk #10 succeeded at 1398 with fuzz 1 (offset 78 lines). 7 out of 10 hunks FAILED patch program exited with status 1
Jeroen, You forgot about -R parameter. This patch is original patch from upstream, so when we want to revert it we should use -R parameter. I've also used this parameter in my ebuild patch attachment. In my test box I have: ~ # cd /usr/src /usr/src # tar xf /usr/portage/distfiles/cacti-spine-0.8.8d.tar.gz /usr/src # cd cacti-spine-0.8.8d/ /usr/src/cacti-spine-0.8.8d # wget "https://552030.bugs.gentoo.org/attachment.cgi?id=405092" -O cacti-spine-0.8.8d-polling_time.patch --2015-06-14 09:34:45-- https://552030.bugs.gentoo.org/attachment.cgi?id=405092 Resolving 552030.bugs.gentoo.org... 204.187.15.4, 2607:fcc0:4:ffff::4 Connecting to 552030.bugs.gentoo.org|204.187.15.4|:443... connected. HTTP request sent, awaiting response... 200 OK Length: 8978 (8.8K) [text/plain] Saving to: ‘cacti-spine-0.8.8d-polling_time.patch’ cacti-spine-0.8.8d-polling_time.patch 100%[=======================================================================================================>] 8.77K --.-KB/s in 0s 2015-06-14 09:34:47 (24.6 MB/s) - ‘cacti-spine-0.8.8d-polling_time.patch’ saved [8978/8978] /usr/src/cacti-spine-0.8.8d # patch -p3 -R <cacti-spine-0.8.8d-polling_time.patch patching file poller.c /usr/src/cacti-spine-0.8.8d #
(In reply to Tomasz Chilinski from comment #9) > Jeroen, You forgot about -R parameter. This patch is original patch from > upstream, so when we want to revert it we should use -R parameter. I've also > used this parameter in my ebuild patch attachment. "One of these patches is not like the others." That sucks.
Those patches are in. And a bit more. Thanks!