Created attachment 179475 [details] Possible fix. This port has the following settings for rcorder: # PROVIDE: dhcp6c # REQUIRE: DAEMON # BEFORE: NETWORKING # KEYWORD: shutdown # PROVIDE: dhcp6s # REQUIRE: DAEMON # BEFORE: NETWORKING # KEYWORD: shutdown This tells rcorder to put dhcp6{c,s} after DAEMON and before NETWORKING. So, eg: DAEMON dhcp6c dhcp6s NETWORKING Unfortunately, this conflicts with DAEMON: # PROVIDE: DAEMON # REQUIRE: NETWORKING SERVERS Which requires that NETWORKING come before DAEMON. The following error is produced by rcorder: rcorder: Circular dependency on provision `DAEMON' in file `/usr/local/etc/rc.d/dhcp6c'. rcorder: Circular dependency on provision `DAEMON' in file `/usr/local/etc/rc.d/dhcp6s'. As a fix I'd suggest just removing the requirement on DAEMON, since running before NETWORKING seems more important. This was introduced in bug #169138, so has been this way for nearly 4 years!
Committed, thanks.
A commit references this bug: Author: pi Date: Sun Sep 2 21:02:11 UTC 2018 New revision: 478819 URL: https://svnweb.freebsd.org/changeset/ports/478819 Log: net/dhcp6: several fixes - rcorder incorrect (Circular dependency on provision `DAEMON') - command_flags do not need to be referenced inside of the command_args - dhcp6s exits because of shutdown on non-connected UDP socket PR: 216673, 217452, 228994 Submitted by: tdb, Andrey Pevnev <apevnev@me.com>, leres, Dries Michiels <driesm.michiels@gmail.com> Approved by: hrs (maintainer timeout) Differential Revision: D16994 Changes: head/net/dhcp6/Makefile head/net/dhcp6/files/dhcp6c.in head/net/dhcp6/files/dhcp6s.in head/net/dhcp6/files/patch-dhcp6s.c
Unless I'm missing something, the change in this bug wasn't actually in the commit?
A commit references this bug: Author: pi Date: Sun Sep 2 21:14:48 UTC 2018 New revision: 478821 URL: https://svnweb.freebsd.org/changeset/ports/478821 Log: net/dhcp6: really fix the rcorder - ups, missed that one PR: 216673 Submitted by: tdb Changes: head/net/dhcp6/files/dhcp6c.in head/net/dhcp6/files/dhcp6s.in
(In reply to Tim Bishop from comment #3) Yes, you're right. Fixed. I used the diff from the review and did not check.