Re: named startup problems upgrading from 7.1p4 to 7.1p5 or 7.1p6
On Sun, 28 Jun 2009 20:54:26 Ian wrote: Hi, I've been meaning to sort this out since the release of 7.1p5, but only just got around to it - I have an installation of 7.1 that runs bind and has been working fine up until I tried to update the system to 7.1p5 (using freebsd-update). As soon as I apply the update reboot, named loads but the startup script hangs. If I press Ctrl+C, the system continues to boot. If I then run /etc/rc.d/named start, named starts, but again the script hangs. I can do DNS lookups while named is running, so it seems to be functioning ok. I tried adding various echo statements to /etc/rc.d/named and found that the script seems to run right through. The hang occurs where /etc/rc.subr echoes out Starting named after the named script has run and that's where things seem to stop! Nothing else that is started by the rc.d scripts hangs, so I'm guessing /etc/rc.subr is ok. I did a diff of /etc/rc.d/named before after the upgrade from p4 to p5 (or p6 which has the same issue) and there are no changes to the file. Nothing seems to be logged anywhere that shows a problem, so I really have no idea what to check next. The only named entry in rc.conf is named_enable=YES. Doing a freebsd-update rollback restores normal operation and given that bind actually loads 7 seems to work apart fromthe hanging script, I suspect there's nothing wrong with my bind configuration. Any suggestions? Cheers, -- Ian gpg key: http://home.swiftdsl.com.au/~imoore/no-spam.asc I've never really solved this problem - even running with the default named.conf as a simple caching server didn't change anything. Instead, I rolled back to 7.1p4 then upgraded to 7.2(p2) and bind works just fine. Cheers, -- Ian gpg key: http://home.swiftdsl.com.au/~imoore/no-spam.asc signature.asc Description: This is a digitally signed message part.
Re: named startup problems upgrading from 7.1p4 to 7.1p5 or 7.1p6
Sorry for starting a new thread with this - my ISP's mail server seems to rejecting all mail recipients when I send email with a mail client, so I'm having to use webmail instead. Their tech says they won't help - they only support Outlook! Grrr! On Sun, 28 Jun 2009 23:27:07 Matthew Seaman wrote: Ian wrote: Well the fact that if I run /etc/rc.d/named manually after the system has booted, the script also hangs suggests it's not the next process I have just check however ntpdate is the next one in the list to be started and that does start correctly - you can see it report the clock being adjusted. Also, when you do a Ctrl+C to break the named script on bootup, it says Script /etc/rc.d/named interrupted. Something I've just realised is that named stays loaded even when you 'break' the script. on bootup and DNS lookups work (I didn't think that was the case originally, but it is). Actually, some careful checking tonight shows that I had forgotten I had a second DNS server in resolv.conf that was doing the DNS resolution - in fact bind on this server is not working even though a bind process appears to be running :/ Hmmm Anything interesting from named in the system logs? You might want to enable /var/log/all.log by following the instructions in /etc/syslog.conf and then see what output you get by bouncing named. It's usually pretty good at pointing out exactly what it thinks the problem is. I've enabled all.log, it only shows the following output when starting named: Jun 29 20:51:43 msgserver named[1593]: starting BIND 9.4.2-P2 -t /var/named -u bind Jun 29 20:51:43 msgserver named[1593]: found 1 CPU, using 1 worker thread a ps axw | grep named gives the following output after running /etc/rc.d/named start: 1988 ?? Is 0:00.00 /usr/sbin/named -t /var/named -u bind 1930 p0 I+ 0:00.06 /bin/sh -x /etc/rc.d/named start 1987 p0 I+ 0:00.01 /usr/sbin/named -t /var/named -u bind and then after doing a Ctrl+C in the terminal where /etc/rc.d/named start is running, only one process continues to run: 1988 ?? Is 0:00.00 /usr/sbin/named -t /var/named -u bind This process doesn't respond to DNS queries, to rndc commands, to /etc/rc.d/named stop (says no process is running because there is no pid file being created) or by a kill command other than kill -9. (All named processes were killed before stating named) You could also try running: # /bin/sh -x /etc/rc.d/named start -- make sure named isn't running when you do that. There will be quite a lot of output as the rc system loads all of the various config files, but you should be able to trace exactly where it's got to when it does hang. Here's the edited highlights of output, I can't see anything that helps: + _rc_subr_loaded=: + name=named + rcvar=named_enable + command=/usr/sbin/named + extra_commands=reload + start_precmd=named_precmd + start_postcmd=make_symlinks + reload_cmd=named_reload + stop_cmd=named_stop + stop_postcmd=named_poststop + load_rc_config named + _name=named + [ -z named ] + false + [ -r /etc/defaults/rc.conf ] snip + named_enable=NO + named_program=/usr/sbin/named + named_pidfile=/var/run/named/pid + named_uid=bind + named_chrootdir=/var/named + named_chroot_autoupdate=YES + named_symlink_enable=YES snip + sourced_files=:/etc/rc.conf::/etc/rc.conf.local: + [ -r /etc/rc.conf.local ] + _rc_conf_loaded=true + [ -f /etc/rc.conf.d/named ] + required_dirs=/var/named + pidfile=/var/run/named/pid + command_args=-u bind + run_rc_command start + _return=0 + rc_arg=start + [ -z named ] + shift 1 + rc_extra_args= + _rc_prefix= + eval _override_command=$named_program + _override_command=/usr/sbin/named + command=/usr/sbin/named + _keywords=start stop restart rcvar reload + rc_pid= + _pidcmd= + _procname=/usr/sbin/named + [ -n /usr/sbin/named ] + [ -n /var/run/named/pid ] + _pidcmd=rc_pid=$(check_pidfile /var/run/named/pid /usr/sbin/named ) + [ -n rc_pid=$(check_pidfile /var/run/named/pid /usr/sbin/named ) ] + _keywords=start stop restart rcvar reload status poll + [ -z start ] + [ -n ] + eval rc_flags=$named_flags + rc_flags= + eval _chdir=$named_chdir _chroot=$named_chroot _nice=$named_nice _user=$named_user _group=$named_group _groups=$named_groups + _chdir= _chroot= _nice= _user= _group= _groups= + [ -n ] + [ -n named_enable -a start != rcvar ] + checkyesno named_enable + eval _value=$named_enable + _value=YES + debug checkyesno: named_enable is set to YES. + return 0 + eval rc_pid=$(check_pidfile /var/run/named/pid /usr/sbin/named ) + check_pidfile /var/run/named/pid /usr/sbin/named + _pidfile=/var/run/named/pid + _procname=/usr/sbin/named + _interpreter= + [ -z /var/run/named/pid -o -z /usr/sbin/named ] + [ ! -f /var/run/named/pid ] + debug pid file (/var/run/named/pid): not readable. + return + rc_pid= + [ start != start ] + eval _cmd=$start_cmd _precmd=$start_precmd
Re: named startup problems upgrading from 7.1p4 to 7.1p5 or 7.1p6
On Tue, 30 Jun 2009 08:35:26 +, no-s...@people.net.au wrote: Sorry for starting a new thread with this - my ISP's mail server seems to rejecting all mail recipients when I Which which reason? send email with a mail client, so I'm having to use webmail instead. Their tech says they won't help - they only support Outlook! Grrr! Can I read this as they don't support proper POP/SMPT? What an ISP... :-( -- Polytropon From Magdeburg, Germany Happy FreeBSD user since 4.0 Andra moi ennepe, Mousa, ... ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
named startup problems upgrading from 7.1p4 to 7.1p5 or 7.1p6
Hi, I've been meaning to sort this out since the release of 7.1p5, but only just got around to it - I have an installation of 7.1 that runs bind and has been working fine up until I tried to update the system to 7.1p5 (using freebsd-update). As soon as I apply the update reboot, named loads but the startup script hangs. If I press Ctrl+C, the system continues to boot. If I then run /etc/rc.d/named start, named starts, but again the script hangs. I can do DNS lookups while named is running, so it seems to be functioning ok. I tried adding various echo statements to /etc/rc.d/named and found that the script seems to run right through. The hang occurs where /etc/rc.subr echoes out Starting named after the named script has run and that's where things seem to stop! Nothing else that is started by the rc.d scripts hangs, so I'm guessing /etc/rc.subr is ok. I did a diff of /etc/rc.d/named before after the upgrade from p4 to p5 (or p6 which has the same issue) and there are no changes to the file. Nothing seems to be logged anywhere that shows a problem, so I really have no idea what to check next. The only named entry in rc.conf is named_enable=YES. Doing a freebsd-update rollback restores normal operation and given that bind actually loads 7 seems to work apart fromthe hanging script, I suspect there's nothing wrong with my bind configuration. Any suggestions? Cheers, -- Ian gpg key: http://home.swiftdsl.com.au/~imoore/no-spam.asc signature.asc Description: This is a digitally signed message part.
Re: named startup problems upgrading from 7.1p4 to 7.1p5 or 7.1p6
Ian wrote: Hi, I've been meaning to sort this out since the release of 7.1p5, but only just got around to it - I have an installation of 7.1 that runs bind and has been working fine up until I tried to update the system to 7.1p5 (using freebsd-update). As soon as I apply the update reboot, named loads but the startup script hangs. If I press Ctrl+C, the system continues to boot. If I then run /etc/rc.d/named start, named starts, but again the script hangs. I can do DNS lookups while named is running, so it seems to be functioning ok. I tried adding various echo statements to /etc/rc.d/named and found that the script seems to run right through. The hang occurs where /etc/rc.subr echoes out Starting named after the named script has run and that's where things seem to stop! Nothing else that is started by the rc.d scripts hangs, so I'm guessing /etc/rc.subr is ok. I did a diff of /etc/rc.d/named before after the upgrade from p4 to p5 (or p6 which has the same issue) and there are no changes to the file. Nothing seems to be logged anywhere that shows a problem, so I really have no idea what to check next. The only named entry in rc.conf is named_enable=YES. Doing a freebsd-update rollback restores normal operation and given that bind actually loads 7 seems to work apart fromthe hanging script, I suspect there's nothing wrong with my bind configuration. Any suggestions? Are you sure it's not the thing which starts immediately *after* named that is hanging? Try running: # rcorder /etc/rc.d/* /usr/local/etc/rc.d/* and see what should come next. Note this command shows the order in which all of the rc scripts in those directories would run, not just the ones you have enabled in rc.conf, so you may well have to skip a few lines until you get to something that is enabled. Cheers, Matthew -- Dr Matthew J Seaman MA, D.Phil. 7 Priory Courtyard Flat 3 PGP: http://www.infracaninophile.co.uk/pgpkey Ramsgate Kent, CT11 9PW signature.asc Description: OpenPGP digital signature
Re: named startup problems upgrading from 7.1p4 to 7.1p5 or 7.1p6
On Sun, 28 Jun 2009 21:43:49 Matthew Seaman wrote: Ian wrote: Hi, I've been meaning to sort this out since the release of 7.1p5, but only just got around to it - I have an installation of 7.1 that runs bind and has been working fine up until I tried to update the system to 7.1p5 (using freebsd-update). As soon as I apply the update reboot, named loads but the startup script hangs. If I press Ctrl+C, the system continues to boot. If I then run /etc/rc.d/named start, named starts, but again the script hangs. I can do DNS lookups while named is running, so it seems to be functioning ok. I tried adding various echo statements to /etc/rc.d/named and found that the script seems to run right through. The hang occurs where /etc/rc.subr echoes out Starting named after the named script has run and that's where things seem to stop! Nothing else that is started by the rc.d scripts hangs, so I'm guessing /etc/rc.subr is ok. I did a diff of /etc/rc.d/named before after the upgrade from p4 to p5 (or p6 which has the same issue) and there are no changes to the file. Nothing seems to be logged anywhere that shows a problem, so I really have no idea what to check next. The only named entry in rc.conf is named_enable=YES. Doing a freebsd-update rollback restores normal operation and given that bind actually loads 7 seems to work apart fromthe hanging script, I suspect there's nothing wrong with my bind configuration. Any suggestions? Are you sure it's not the thing which starts immediately *after* named that is hanging? Try running: # rcorder /etc/rc.d/* /usr/local/etc/rc.d/* and see what should come next. Note this command shows the order in which all of the rc scripts in those directories would run, not just the ones you have enabled in rc.conf, so you may well have to skip a few lines until you get to something that is enabled. Cheers, Matthew Well the fact that if I run /etc/rc.d/named manually after the system has booted, the script also hangs suggests it's not the next process I have just check however ntpdate is the next one in the list to be started and that does start correctly - you can see it report the clock being adjusted. Also, when you do a Ctrl+C to break the named script on bootup, it says Script /etc/rc.d/named interrupted. Something I've just realised is that named stays loaded even when you 'break' the script. on bootup and DNS lookups work (I didn't think that was the case originally, but it is). Cheers, -- Ian gpg key: http://home.swiftdsl.com.au/~imoore/no-spam.asc signature.asc Description: This is a digitally signed message part.
Re: named startup problems upgrading from 7.1p4 to 7.1p5 or 7.1p6
Ian wrote: Well the fact that if I run /etc/rc.d/named manually after the system has booted, the script also hangs suggests it's not the next process I have just check however ntpdate is the next one in the list to be started and that does start correctly - you can see it report the clock being adjusted. Also, when you do a Ctrl+C to break the named script on bootup, it says Script /etc/rc.d/named interrupted. Something I've just realised is that named stays loaded even when you 'break' the script. on bootup and DNS lookups work (I didn't think that was the case originally, but it is). Hmmm Anything interesting from named in the system logs? You might want to enable /var/log/all.log by following the instructions in /etc/syslog.conf and then see what output you get by bouncing named. It's usually pretty good at pointing out exactly what it thinks the problem is. You could also try running: # /bin/sh -x /etc/rc.d/named start -- make sure named isn't running when you do that. There will be quite a lot of output as the rc system loads all of the various config files, but you should be able to trace exactly where it's got to when it does hang. You're using the system-supplied copy of bind aren't you? Have you got a valid /etc/named/rndc.conf or /etc/named/rndc.key file so you can use rndc(8)? If not, try running: # rndc-confgen /etc/namedb/rndc.conf and then cut'n'paste the indicated key and controls statements from that file into named.conf, stripping out the comment characters as you do (of course). If you're using one of the ports versions of named, do exactly the same thing, but copy or link rndc.conf into /usr/local/etc/ as well. Cheers, Matthew -- Dr Matthew J Seaman MA, D.Phil. 7 Priory Courtyard Flat 3 PGP: http://www.infracaninophile.co.uk/pgpkey Ramsgate Kent, CT11 9PW signature.asc Description: OpenPGP digital signature
Re: named startup problems upgrading from 7.1p4 to 7.1p5 or 7.1p6
On Sunday 28 June 2009 03:24:26 Ian wrote: I tried adding various echo statements to /etc/rc.d/named and found that the script seems to run right through. rc_debug=YES in /etc/rc.conf is REALLY handy for this. -- Mel ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
