syslogd after named at boot

My FreeBSD 8.2-Stable system is starting syslogd before named. This results in syslogd failing. I get:
Code:
syslogd: bind: Can't assign requested address
Once the system is up, I can log in and start syslogd manually.

The problem is somewhat described in this this thread I say somewhat because named isn't what caused that issue.

This system was just csup'ed from 8.0-Stable (csup; make buildworld etc.). Before that, IIRC it was updated by source from 7.x. The problem existed in 8.0, but I ignored it. I'm sure I screwed up something I needed to merge in via mergemaster at that time.

I'm looking for a clue as to how to fix things now, if possible. All my custom changes are (should be) in /etc/rc.conf only.

I tried teaching /etc/rc.d/syslogd to require named, it didn't help. But I might not have done the right thing. I'm only just looking at how all this works.

Any suggestions?

Thanks


Mod Edit: Solution listed in the following PR: ports/152969. See post 25 below for more details.
 
syslogd doesn't require named so the order is not important. The reason syslogd isn't starting is related to something else.

Try setting this:
Code:
syslogd_flags="-ss"

That will prevent syslogd from opening a port, which you probably don't need anyway.
 
SirDice said:
syslogd doesn't require named so the order is not important. The reason syslogd isn't starting is related to something else.

Try setting this:
Code:
syslogd_flags="-ss"

That will prevent syslogd from opening a port, which you probably don't need anyway.


I'll try that to get past this issue.

Does syslogdtry to resolve a DNS name to IP address? I searched here and via Google, and came up with named as a common symptom.

I'll probably need syslogd to log over the net on that system as well, so I wouldn't mind finding out what's going on. After the system is up, starting syslogd with the same settings 'just works'

Thanks
 
Is your /etc/hosts intact after mergemaster? It appears that syslogd is having trouble binding to interfaces like localhost (which, like all local hostnames, should always be present in hosts to make everything work in the absence of resolving).
 
DutchDaemon said:
Is your /etc/hosts intact after mergemaster? It appears that syslogd is having trouble binding to interfaces like localhost (which, like all local hostnames, should always be present in hosts to make everything work in the absence of resolving).

They look right. Could it be that the lo0 interface isn't there yet?


Code:
Oct 17 19:49:17 wan2 kernel: Creating and/or trimming log files
Oct 17 19:49:17 wan2 kernel: .
Oct 17 19:49:17 wan2 kernel: Starting syslogd.
Oct 17 19:49:17 wan2 kernel: Oct 17 15:47:50  syslogd: bind: Can't assign reques
ted address
Oct 17 19:49:17 wan2 kernel: Oct 17 15:47:50  syslogd: bind: Can't assign reques
ted address
Oct 17 19:49:17 wan2 kernel: syslogd:
Oct 17 19:49:17 wan2 kernel: child pid 178 exited with return code 1
Oct 17 19:49:17 wan2 kernel:
Oct 17 19:49:17 wan2 kernel: /etc/rc: WARNING: failed to start syslogd
Oct 17 19:49:17 wan2 kernel: Starting Network: em0 xl0 lo0.

syslogd fails, then Starting Network: em0 xl0 lo0

Still digging, thought I'd update here with what I've found so far.
 
Yeah, that doesn't look right. I checked about a dozen dmesg logs I have lying around, and they all start their network interfaces well and truly before syslogd shows up. Are you on DHCP, perhaps? If so, can you try "SYNCDHCP" instead of "DHCP" in /etc/rc.conf?
 
DutchDaemon said:
Yeah, that doesn't look right. I checked about a dozen dmesg logs I have lying around, and they all start their network interfaces well and truly before syslogd shows up. Are you on DHCP, perhaps? If so, can you try "SYNCDHCP" instead of "DHCP" in /etc/rc.conf?

I do have DHCP configured, but am not using it. That is, the xl0 interface will use DHCP when I plug the cable in. em0 has its IPv4 (and IPv6 iirc) statically assigned in rc.conf.

I'll try SYNCDHCP tonight.

Thanks
 
8-STABLE and 9.0 have /etc/rc.d/netwait for static interfaces.

Untested...
/etc/rc.conf
Code:
netwait_enable="YES"
netwait_ip="192.168.1.1"  # network router to attempt to ping
 
wblock@ said:
8-STABLE and 9.0 have /etc/rc.d/netwait for static interfaces.

Thanks, I'll try that too. This machine will use DHCP to my ISP, once I move the configuration over off the current (blush) 4.10-Stable system which served me so well. The HW is on its way out.

When done, DHCP will be used for the ISP link... static IP for the private IP subnets. That said, will using /etc/rc.d/netwait still make sense?
 
BostonDriver said:
When done, DHCP will be used for the ISP link... static IP for the private IP subnets. That said, will using /etc/rc.d/netwait still make sense?

Probably not. It would be useful in a situation where there's a static IP with the ISP but the card takes a while to come up. The equivalent of SYNCDHCP, but with a static address. If this system is a router, all the systems inside the LAN are waiting on it, instead of the other way around.
 
DutchDaemon said:
Yeah, that doesn't look right. I checked about a dozen dmesg logs I have lying around, and they all start their network interfaces well and truly before syslogd shows up. Are you on DHCP, perhaps? If so, can you try "SYNCDHCP" instead of "DHCP" in /etc/rc.conf?

SYNCDHCP didn't help:

Code:
[mcc@wan2 /etc]$ grep ifconfig rc.conf
#ifconfig_xl0="DHCP link 00:60:97:29:65:71"
ifconfig_xl0="SYNCDHCP link 00:60:97:29:65:71"
ifconfig_em0="inet 172.16.8.2  netmask 255.255.255.0"

Code:
Oct 18 19:25:26 wan2 kernel: Creating and/or trimming log files
Oct 18 19:25:26 wan2 kernel: .
Oct 18 19:25:26 wan2 kernel: Starting syslogd.
Oct 18 19:25:26 wan2 kernel: Oct 18 15:24:43  syslogd: bind: Can't assign reques
ted address
Oct 18 19:25:26 wan2 kernel: Oct 18 15:24:43  syslogd: bind: Can't assign reques
ted address
Oct 18 19:25:26 wan2 kernel: syslogd:
Oct 18 19:25:26 wan2 kernel: child pid 181 exited with return code 1
Oct 18 19:25:26 wan2 kernel:
Oct 18 19:25:26 wan2 kernel: /etc/rc: WARNING: failed to start syslogd
Oct 18 19:25:26 wan2 kernel: xl0: no link ...
Oct 18 19:25:26 wan2 kernel: .
Oct 18 19:25:26 wan2 last message repeated 10 times
Oct 18 19:25:26 wan2 kernel: giving up
Oct 18 19:25:26 wan2 kernel: Starting Network: lo0 em0 xl0.
 
I took link out, and just used SYNCDHCP... same problem.

I removed DHCP, and supplied static IPv4 addresses for xl0, same problem.

Time for a pint.
 
Please show the whole rc.conf. Wouldn't hurt to show /etc/hosts, too.

Anything weird going on? Jails, edits of /etc/defaults/rc.conf?
 
wblock@ said:
Please show the whole rc.conf. Wouldn't hurt to show /etc/hosts, too.

Anything weird going on? Jails, edits of /etc/defaults/rc.conf?


No jails I've created. Maybe e.g. named (other subsystems) do on my behalf?

No mods to /etc/defaults/rc.conf, though a mergemaster mistake isn't out of the question.

I cut some noise out of rc.conf As it currently is, here is the data:

/var/log/console.log

Code:
Oct 18 22:39:30 wan2 kernel: Setting hostname: wan2.fid4.com
Oct 18 22:39:30 wan2 kernel: .
Oct 18 22:39:30 wan2 kernel: Creating and/or trimming log files
Oct 18 22:39:30 wan2 kernel: .
Oct 18 22:39:30 wan2 kernel: Starting syslogd.
Oct 18 22:39:30 wan2 kernel: Oct 18 18:38:17  syslogd: bind: Can't assign reques
ted address
Oct 18 22:39:30 wan2 kernel: Oct 18 18:38:17  syslogd: bind: Can't assign reques
ted address
[color="Red"]Oct 18 22:39:30 wan2 kernel: syslogd:
Oct 18 22:39:30 wan2 kernel: child pid 174 exited with return code 1[/color]
Oct 18 22:39:30 wan2 kernel:
Oct 18 22:39:30 wan2 kernel: /etc/rc: WARNING: failed to start syslogd
Oct 18 22:39:30 wan2 kernel: Starting Network: lo0 em0 xl0.
Oct 18 22:39:30 wan2 kernel: lo0: flags=8049<UP,LOOPBACK,RUNNING,MULTICAST> metr
ic 0 mtu 16384

hosts (comments deleted.)

Code:
::1                     localhost localhost.fid4.com
127.0.0.1               localhost localhost.fid4.com

/etc/rc.conf

Code:
[mcc@wan2 ~]$ cat /etc/rc.conf

# -- sysinstall generated deltas -- # Sat Sep 29 10:30:57 2007
# Created: Sat Sep 29 10:30:57 2007
# Enable network daemons for user convenience.
# Please make all changes to this file, not to /etc/defaults/rc.conf.
# This file now contains just the overrides from /etc/defaults/rc.conf.
#
network_interfaces="lo0 em0 xl0 gif0"
#
defaultrouter="172.16.8.1"
#
gateway_enable="YES"
inetd_enable="YES"
moused_enable="YES"
moused_flags=""
moused_port="/dev/psm0"
moused_type="auto"
nfs_client_enable="YES"
nfs_server_enable="YES"
rpcbind_enable="YES"
sshd_enable="YES"
saver="daemon"
#
#ifconfig_xl0="DHCP link 00:60:97:29:65:71"
#ifconfig_xl0="SYNCDHCP"
#ifconfig_xl0="DHCP"
ifconfig_xl0="inet 172.16.4.1  netmask 255.255.255.0"
#
#ifconfig_em0="DHCP"
ifconfig_em0="inet 172.16.8.2  netmask 255.255.255.0"
#
#
ipv6_enable="YES"
#ipv6_enable="NO"
#
ipv6_gateway_enable="YES"
rtadvd_enable="YES"
rtadvd_interfaces="em0"
#
ipv6_network_interfaces="lo0 em0 xl0"
ipv6_ifconfig_em0="2001:05c0:84a7:8::1 prefixlen 64"
#
hostname="wan2.fid4.com"
#rc_debug="YES"
#rc_info="YES"
router_enable="YES"
router="/sbin/routed"
router_flags="-s"
#
named_enable="YES"
named_wait="YES"
#
#
cupsd_enable="YES"
#
fusefs_enable="YES"
#
linux_enable="YES"
#
dbus_enable="YES"
#polkitd_enable="YES"
hald_enable="YES"
#
#freenet6_enable="YES"
#
mysql_enable="YES"
#
natd_enable="YES"
natd_interface="xl0"
natd_flags="-dynamic -use_sockets -same_ports -log_ipfw_denied"
#
syslogd_enable="YES"
syslogd_flags="-s"
#
iked_enable="NO"
spmd_enable="NO"
#

Notice the log shows syslogd failed.

Using the exact same rc.conf, with all ipv6 comment out, I still get syslogd logging that it cannot assign the requested address, but syslogd itself does NOT fail. It is running when I log in.
 
That's an eclectic mix of services. Not anything obvious to me, but I wonder if inetd is starting something that's supposed to be done by /etc/rc.d. Is there a reason to set network_interfaces instead of letting it default? Also, the syslogd values are default anyway.
 
wblock@ said:
That's an eclectic mix of services. Not anything obvious to me, but I wonder if inetd is starting something that's supposed to be done by /etc/rc.d. Is there a reason to set network_interfaces instead of letting it default? Also, the syslogd values are default anyway.

I timed that rc.conf down to cut down on what isn't causing the problem ;)

All those services are in the 4.10-Stable system currently running that I hope this system will replace.

I'll try removing network_interfaces and see what happens.

I'll also start rc.conf from scratch over the weekend... adding changes until I find what seems to turn this on.

Thanks
 
Any chance the key to this is multiple NICs?

I have an 8.2-Stable system I installed from DVD (FreeBSD torrent) that has the same issue described above where syslogd fails to bind, but does not crash. rc.conf has IPv6 disabled. (I'll enable next chance I get and see if syslogd crashes)

I have another 8.1 system, from iso, running in a KVM (Fedora 13) that works perfectly. It has just one (emulated) NIC. Moving this working rc.conf from the virtual machine to the other systems results in the problem. If IPv6 enabled, syslogd crashes. If IPv6 disabled, syslogd fails to bind but doesn't crash.

I'll re-test to confirm, taking notes, but wanted to throw this out in case it might connect some dots.
 
I've never had any issues with syslogd failing to bind/start on machines with 1-4 NICs, with or without DHCP, with cloned interfaces, bridges, vlans, DMZs, etc. etc.
 
DutchDaemon said:
I've never had any issues with syslogd failing to bind/start on machines with 1-4 NICs, with or without DHCP, with cloned interfaces, bridges, vlans, DMZs, etc. etc.


I setup my KVM FreeBSD guest with 2 (emulated) NICs and
Code:
ipv6_enable="YES"
Everything behaved. Both NICs were the same driver em0 and em1, in case it makes a difference, but I doubt it.

On the 64bit machine, with a rather minimal rc.conf, running 8.2-Stable from 2011 March, I set
Code:
rc_debug="YES"
in rc.conf and have the results. Even though syslogd doesn't start at boot, once I manually start it, the (buffered) console data is put in /var/log/console.log It clearly shows syslogd being started very early in the boot process.

I'll post the boot through network entries next.
 
Code:
Oct 21 17:31:41 p690 kernel: /etc/rc.shutdown: DEBUG: run_rc_command: doit: sync
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: hostid_enable is set to YES.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: doit: hostid_start
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: rc_startmsgs is set to YES.
Oct 21 17:36:03 p690 kernel: Setting hostuuid: 44454c4c-5000-1037-8046-b1c04f434331.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: rc_startmsgs is set to YES.
Oct 21 17:36:03 p690 kernel: Setting hostid: 0x01af23c5.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: zfs_enable is set to NO.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: doit: dumpon_start
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: ddb_enable is set to NO.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: doit: initrandom_start
Oct 21 17:36:03 p690 kernel: Entropy harvesting:
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: harvest_interrupt is set to YES.
Oct 21 17:36:03 p690 kernel: interrupts
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: harvest_ethernet is set to YES.
Oct 21 17:36:03 p690 kernel: ethernet
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: harvest_p_to_p is set to YES.
Oct 21 17:36:03 p690 kernel: point_to_point
Oct 21 17:36:03 p690 kernel: kickstart
Oct 21 17:36:03 p690 kernel: .
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: start_precmd: [ -n "$(geli_make_list)" ]
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: start_precmd: find_gbde_devices start
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: doit: gbde_start
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: doit: encswap_attach
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: doit: ccd_start
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: doit: swapon -aq
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: doit: fsck_start
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: rc_startmsgs is set to YES.
Oct 21 17:36:03 p690 kernel: Starting file system checks:
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: background_fsck is set to YES.
Oct 21 17:36:03 p690 kernel: /dev/da0s2a: FILE SYSTEM CLEAN; SKIPPING CHECKS
Oct 21 17:36:03 p690 kernel: /dev/da0s2a: clean, 114097 free (2705 frags, 13924 blocks, 0.5% fragmentation)
Oct 21 17:36:03 p690 kernel: /dev/da0s2e: FILE SYSTEM CLEAN; SKIPPING CHECKS
Oct 21 17:36:03 p690 kernel: /dev/da0s2e: clean, 506473 free (41 frags, 63304 blocks, 0.0% fragmentation)
Oct 21 17:36:03 p690 kernel: /dev/da0s2f: FILE SYSTEM CLEAN; SKIPPING CHECKS
Oct 21 17:36:03 p690 kernel: /dev/da0s2f: clean, 84490467 free (84947 frags, 10550690 blocks, 0.1% fragmentation)
Oct 21 17:36:03 p690 kernel: /dev/da0s2d: FILE SYSTEM CLEAN; SKIPPING CHECKS
Oct 21 17:36:03 p690 kernel: /dev/da0s2d: clean, 4980872 free (1400 frags, 622434 blocks, 0.0% fragmentation)
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: doit: root_start
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: hostid_enable is set to YES.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: doit: hostid_save
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: start_precmd: [ -n "${_mdconfig_list}" ]
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: doit: mountcritlocal_start
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: rc_startmsgs is set to YES.
Oct 21 17:36:03 p690 kernel: Mounting local file systems:
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: rc_startmsgs is set to YES.
Oct 21 17:36:03 p690 kernel: .
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: zfs_enable is set to NO.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: cleanvar_enable is set to YES.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: start_precmd: cleanvar_prestart
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: doit: cleanvar_start
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: doit: devfs_start
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: ipfilter_enable is set to NO.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: doit: addswap_start
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: doit: auto_linklocal_start
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: ipv6_enable is set to YES.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: doit: sysctl_start
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: doit: hostname_start
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: rc_startmsgs is set to YES.
Oct 21 17:36:03 p690 kernel: Setting hostname: p690.fid4.com
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: rc_startmsgs is set to YES.
Oct 21 17:36:03 p690 kernel: .
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: ipmon_enable is set to NO.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: start_precmd: [ -n "${_mdconfig2_list}" ]
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: newsyslog_enable is set to YES.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: doit: newsyslog_start
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: rc_startmsgs is set to YES.
Oct 21 17:36:03 p690 kernel: Creating and/or trimming log files
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: rc_startmsgs is set to YES.
Oct 21 17:36:03 p690 kernel: .
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: pid file (/var/run/syslog.pid): not readable.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: syslogd_enable is set to YES.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: start_precmd: syslogd_precmd
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: named_enable is set to NO.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: rc_startmsgs is set to YES.
Oct 21 17:36:03 p690 kernel: Starting syslogd.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: doit: /usr/sbin/syslogd -s
Oct 21 17:36:03 p690 kernel: Oct 21 13:33:34  syslogd: bind: Can't assign requested address
Oct 21 17:36:03 p690 kernel: Oct 21 13:33:34  syslogd: bind: Can't assign requested address
Oct 21 17:36:03 p690 kernel: syslogd:
Oct 21 17:36:03 p690 kernel: child pid 203 exited with return code 1
Oct 21 17:36:03 p690 kernel:
Oct 21 17:36:03 p690 kernel: /etc/rc: WARNING: failed to start syslogd
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: start_precmd: savecore_prestart
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: No dump device. Quitting.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: doit: ldconfig_start
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: ldconfig_insecure is set to NO.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: rc_startmsgs is set to YES.
Oct 21 17:36:03 p690 kernel: ELF ldconfig path: /lib /usr/lib /usr/lib/compat /usr/local/lib /usr/local/lib/compat/pkg /usr/local/kde4/lib /usr/local/
lib /usr/local/lib/compat/pkg /usr/local/lib/mysql /usr/local/lib/pth /usr/local/lib/qt4 /usr/local/lib/speech-dispatcher
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: rc_startmsgs is set to YES.
Oct 21 17:36:03 p690 kernel: 32-bit compatibility ldconfig path:
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: doit: archdep_start
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: doit: abi_start
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: sysvipc_enable is set to NO.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: linux_enable is set to YES.
Oct 21 17:36:03 p690 kernel: Additional ABI support:
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: sysvipc_enable is set to NO.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: linux_enable is set to YES.
Oct 21 17:36:03 p690 kernel: linux
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: load_kld: linux kernel module already loaded.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: svr4_enable is set to NO.
Oct 21 17:36:03 p690 kernel: .
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: pid file (/var/run/named/pid): not readable.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: named_enable is set to NO.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: doit: random_start
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: doit: adjkerntz -i
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: atm_enable is set to NO.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: kldxref_enable is set to NO.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: doit: sppp_start
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: ipnat_enable is set to NO.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: ipfs_enable is set to NO.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: run_rc_command: doit: network_start
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: Cloned:
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: synchronous_dhclient is set to NO.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: synchronous_dhclient is set to NO.
Oct 21 17:36:03 p690 kernel: Starting Network: lo0 em0 bge0.
Oct 21 17:36:03 p690 kernel: /etc/rc: DEBUG: checkyesno: rc_startmsgs is set to YES.
Oct 21 17:36:03 p690 kernel: lo0: flags=8049<UP,LOOPBACK,RUNNING,MULTICAST> metric 0 mtu 16384
Oct 21 17:36:03 p690 kernel: options=3<RXCSUM,TXCSUM>
Oct 21 17:36:03 p690 kernel: inet 127.0.0.1 netmask 0xff000000
 
Here is the rc.conf Not much in this case.

Code:
# -- sysinstall generated deltas -- # Sun Mar 13 10:00:06 2011
# Created: Sun Mar 13 10:00:06 2011
# Enable network daemons for user convenience.
# Please make all changes to this file, not to /etc/defaults/rc.conf.
# This file now contains just the overrides from /etc/defaults/rc.conf.
rc_debug="YES"
gateway_enable="YES"
hostname="p690.fid4.com"
ifconfig_bge0="DHCP"
ifconfig_em0="DHCP"
inetd_enable="YES"
#ipv6_enable="NO"
ipv6_enable="YES"
moused_enable="YES"
nfs_client_enable="YES"
nfs_server_enable="YES"
rpcbind_enable="YES"
saver="daemon"
sshd_enable="YES"
#
dbus_enable="YES"
hald_enable="YES"
#
linux_enable="YES"
#
sem_load="YES"
freenet6_enable="YES"
mysql_enable="YES"
#
 
I found something

On a scratch disk, I installed 8.2-Release from DVD. I mounted the 8.2-Stable disks and diff'ed. /etc/rc.d are virtually identical. Doing similar type of diff's I pretty much ruled out /etc as being where the issue is. So I moved on to /usr/local/etc.

I moved /usr/local/etc/rc.d for safe keeping, created an empty rc.d. The problem described in this post is gone. syslogd starts were I expect it to.

/usr/local/etc/rc.d will have old stuff in it. When the update from 7.x to 8.x hit this boot time issue, I gave up on the upgrade. So none of the e.g. ports were ever updated, as I thought there was a problem I caused in the base system.

Later today, I'll start digging into what specifically might be the cause.

Thanks again for everyone's time.
 
Wow, I've come a full circle.

The problem is described in this thread I linked to in my original post. Even with freenet6 not set in rc.conf, just having the script exist in /usr/local/etc/rc.d/freenet6 results in syslogd starting before the network for some reason.

I foolishly ruled out freenet6, and the specifics in that thread, because I thought it was not enabled.

That thread doesn't have a solution. I'll Google a bit to see if there is a known solution. I also will start to update my ports and packages, starting with freenet6, when I get a chance. I'm out of weekend.
 
Compare the output of $ rcorder /etc/rc.d/* /usr/local/etc/rc.d/* with and without freenet6 enabled, and with and without freenet6 script included in the rc.d directory. Check where in the list syslogd appears, especially in relation to freenet6 and NETWORKING.

If having freenet6 enabled or present in the list moves syslogd in the rcorder output, then look at the freenet6 script. Check the first few lines (requires, before, after, etc). That's probably where the error is.
 
I found this earlier today. It looks like a known issue. Here is the PR I verified the fix just a few seconds ago.

phoenix said:
If having freenet6 enabled or present in the list moves syslogd in the rcorder output, then look at the freenet6 script. Check the first few lines (requires, before, after, etc). That's probably where the error is.

Just having freenet6 exist in /usr/local/etc/rc.d caused the problem for me. When I purged rc.conf of most everything, leaving freenet6 out, I still had the issue (even though freenet6 did not try to start.) That's what caused the time sink.. sent me down a rat hole.

Note: I just updated ports using portsnap fetch update on the 8.2-Stable system csup done a few days ago. The port for freenet6 still doesn't have the fix.

Thanks for everyone's help.

Mods: Perhaps this PR can be mentioned in the link I referenced in my first post? There is at least one other post I found on this forum that mentioned a similar problem. Maybe that post too?
 
Back
Top