Re: Message about ntp when updating
Chris Davies wrote, on 01/12/12 00:12: Jörg-Volker Peetz jvpe...@web.de wrote: Here I'm using rdate -an in a cron job with something like the following rdate -acnv $NTPHOST Just like ntpdate or any of the other cron based solutions proposed in this thread, this does not train the clock. So although rdate slews the clock, it doesn't train it to become more accurate over time. NTP does all this out of the box - and you get multiple server resiliency from the NTP Pool. (Do you really need the -c flag to rdate? My systems don't.) Chris Now that you mentioned it, I checked again for the -c option. Indeed, it screws the clock by approx. 23 sec. (using de.pool.ntp.org). As the man page for rdate states, comparison with a radio controlled clock revealed this. Thank you very much for the hint. I'm using rdate because it's a small uncomplicated (nearly :-) package. And yes, calling rdate every 8 hours I see an approx. constant adjustment value in the log file. -- Best regards, Jörg-Volker. -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/jem823$q16$1...@dough.gmane.org
Message about ntp when updating
Hi, all: I see this message when running an update: Installing new version of config file /etc/cron.daily/ntp ... insserv: warning: current stop runlevel(s) (0 1 6) of script `ntp' overwrites defaults (empty). Should I be concerned about it? And, if so, what am I supposed to do to correct it. Cheers Patrick -- 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/cajvvksodwh6uivswlmbfbf7ceufzvo7uk8fkacthxjqaxy1...@mail.gmail.com
Re: Message about ntp when updating
Good time of the day, Patrick. You worte: I see this message when running an update: Installing new version of config file /etc/cron.daily/ntp ... insserv: warning: current stop runlevel(s) (0 1 6) of script `ntp' overwrites defaults (empty). Should I be concerned about it? And, if so, what am I supposed to do to correct it. Not answering to Your question, but as mine opinion - if You do not intend other hosts to be sync-ing time w/ this host, I think You need the service at all. For sync-ing time from Internet for the host You can use ntpdate in cron/anacron for example. -- 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/4f0da805.c935cc0a.1efe.3...@mx.google.com
Re: Message about ntp when updating
Patrick Wiseman wrote: I see this message when running an update: Installing new version of config file /etc/cron.daily/ntp ... insserv: warning: current stop runlevel(s) (0 1 6) of script `ntp' overwrites defaults (empty). That is a minor upgrade bug in the ntp package. You had a previous version of it (Lenny) installed and have since upgraded to a newer version (Squeeze). The older version included shutdown links so that ntp would be stopped on shutdown. The new package does not include those links anymore. But because they were there in /etc/ they are accidentally preserved after the upgrade when they should have been transitioned when the package was upgraded. For reasons that I haven't been following there has been a push to remove those shutdown links generally from everything. I don't know why. Perhaps someone else will comment on why. Removing those links seems like it breaks going to single user mode directly from multiuser mode but I only go to single user mode from a reboot anyway so I guess that is okay. I am sure the reasoning is that the reboot or halt will kill everything no matter what so might as well get to it as quickly as possible. On a server machine this wouldn't ever be an issue but waiting for a laptop to shutdown could be annoying. The Lenny version had declared: # Default-Start: 2 3 4 5 # Default-Stop:0 1 6 And with links created with: update-rc.d ntp defaults 23 The Squeeze version has declared: # Default-Start: 2 3 4 5 # Default-Stop: And with links created with: update-rc.d ntp start 23 2 3 4 5 . But also remember that Squeeze will use 'insserv' to set up dependency based boot symlinks if possible, ignoring the start 23 2 3 4 5 . part of the above but using the Default-Start/Default-Stop headers instead. In the newer package there are no Default-Stop runlevels declared. Since the previous version had created kill symlinks to take action at shutdown they were left behind. The package upgrade runs the update-rc.d line to set up symlinks and it is producing the message. You can safely try this to see the message in detail: # update-rc.d ntp start 23 2 3 4 5 . update-rc.d: using dependency based boot sequencing insserv: warning: current stop runlevel(s) (0 1 6) of script `ntp' overwrites defaults (empty). That is the type of error that we are going to see a lot of for several different packages as the push to remove shutdown links continues and doesn't clean up after itself. Should I be concerned about it? And, if so, what am I supposed to do to correct it. It is okay to ignore since the issue is for the most part cosmetic. However you might as well clean it up so that it the error message goes away and doesn't obscure a real error message. Find them with: $ find /etc/rc?.d -name 'K*ntp*' /etc/rc0.d/K02ntp /etc/rc1.d/K02ntp /etc/rc6.d/K02ntp Clean them up by removing those files. If you just ran find and are happy with the files listed then you can add -delete to the find command and it will delete them. # find /etc/rc?.d -name 'K*ntp*' -delete Or remove them with an explicit command. # rm -f /etc/rc0.d/K02ntp /etc/rc1.d/K02ntp /etc/rc6.d/K02ntp And then the update-rc.d from the upgrade script will run cleanly without emitting that message. # update-rc.d ntp start 23 2 3 4 5 . update-rc.d: using dependency based boot sequencing Bob signature.asc Description: Digital signature
Re: Message about ntp when updating
Sthu Deus wrote: Not answering to Your question, but as mine opinion - if You do not intend other hosts to be sync-ing time w/ this host, I think You need the service at all. Time is important to most internet hosts. We always hate to see email from users with a date of 1970 for example. I would always install the ntp package to ensure that time is updated on the local host machine. For sync-ing time from Internet for the host You can use ntpdate in cron/anacron for example. Please no. Syncing time from cron is very bad. It causes the clock to be stepped and can create very hard to debug behavior. Setting the clock from cron is the worst way to do this. For example if a process is set to trigger at a particular time and the cron task changes the time out from under it then it will either trigger early or late or possibly not at all. For example on a file server the time is used to set timestamps of files which is used by the make program and jittery clocks can cause all kinds of strange make behavior. And those are just a couple of examples of weird behavior that is possible. It is much better to use a smart time protocol aware daemon such as ntpd which knows how to smoothly adjust the time so that every clock tick is seen and the time is kept updated without clock steps. Bob signature.asc Description: Digital signature
Re: Message about ntp when updating
On Wed, Jan 11, 2012 at 12:26 PM, Bob Proulx b...@proulx.com wrote: Patrick Wiseman wrote: I see this message when running an update: Installing new version of config file /etc/cron.daily/ntp ... insserv: warning: current stop runlevel(s) (0 1 6) of script `ntp' overwrites defaults (empty). That is a minor upgrade bug in the ntp package. You had a previous version of it (Lenny) installed and have since upgraded to a newer version (Squeeze). The older version included shutdown links so that ntp would be stopped on shutdown. The new package does not include those links anymore. But because they were there in /etc/ they are accidentally preserved after the upgrade when they should have been transitioned when the package was upgraded. For reasons that I haven't been following there has been a push to remove those shutdown links generally from everything. I don't know why. Perhaps someone else will comment on why. Removing those links seems like it breaks going to single user mode directly from multiuser mode but I only go to single user mode from a reboot anyway so I guess that is okay. I am sure the reasoning is that the reboot or halt will kill everything no matter what so might as well get to it as quickly as possible. On a server machine this wouldn't ever be an issue but waiting for a laptop to shutdown could be annoying. The Lenny version had declared: # Default-Start: 2 3 4 5 # Default-Stop: 0 1 6 And with links created with: update-rc.d ntp defaults 23 The Squeeze version has declared: # Default-Start: 2 3 4 5 # Default-Stop: And with links created with: update-rc.d ntp start 23 2 3 4 5 . But also remember that Squeeze will use 'insserv' to set up dependency based boot symlinks if possible, ignoring the start 23 2 3 4 5 . part of the above but using the Default-Start/Default-Stop headers instead. In the newer package there are no Default-Stop runlevels declared. Since the previous version had created kill symlinks to take action at shutdown they were left behind. The package upgrade runs the update-rc.d line to set up symlinks and it is producing the message. You can safely try this to see the message in detail: # update-rc.d ntp start 23 2 3 4 5 . update-rc.d: using dependency based boot sequencing insserv: warning: current stop runlevel(s) (0 1 6) of script `ntp' overwrites defaults (empty). That is the type of error that we are going to see a lot of for several different packages as the push to remove shutdown links continues and doesn't clean up after itself. Should I be concerned about it? And, if so, what am I supposed to do to correct it. It is okay to ignore since the issue is for the most part cosmetic. However you might as well clean it up so that it the error message goes away and doesn't obscure a real error message. Find them with: $ find /etc/rc?.d -name 'K*ntp*' /etc/rc0.d/K02ntp /etc/rc1.d/K02ntp /etc/rc6.d/K02ntp Clean them up by removing those files. If you just ran find and are happy with the files listed then you can add -delete to the find command and it will delete them. # find /etc/rc?.d -name 'K*ntp*' -delete Or remove them with an explicit command. # rm -f /etc/rc0.d/K02ntp /etc/rc1.d/K02ntp /etc/rc6.d/K02ntp And then the update-rc.d from the upgrade script will run cleanly without emitting that message. # update-rc.d ntp start 23 2 3 4 5 . update-rc.d: using dependency based boot sequencing Thank you for this above-and-beyond extraordinarily helpful reply! Patrick -- 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/cajvvksnuz6n_o5arrjko-ogawa634zppa7n-jpmnyz+7fcz...@mail.gmail.com
Re: Message about ntp when updating
Bob Proulx wrote, on 01/11/12 18:32: Sthu Deus wrote: Not answering to Your question, but as mine opinion - if You do not intend other hosts to be sync-ing time w/ this host, I think You need the service at all. Time is important to most internet hosts. We always hate to see email from users with a date of 1970 for example. I would always install the ntp package to ensure that time is updated on the local host machine. For sync-ing time from Internet for the host You can use ntpdate in cron/anacron for example. Please no. Syncing time from cron is very bad. It causes the clock to be stepped and can create very hard to debug behavior. Setting the clock from cron is the worst way to do this. For example if a process is set to trigger at a particular time and the cron task changes the time out from under it then it will either trigger early or late or possibly not at all. For example on a file server the time is used to set timestamps of files which is used by the make program and jittery clocks can cause all kinds of strange make behavior. And those are just a couple of examples of weird behavior that is possible. It is much better to use a smart time protocol aware daemon such as ntpd which knows how to smoothly adjust the time so that every clock tick is seen and the time is kept updated without clock steps. Bob Here I'm using rdate -an in a cron job with something like the following rdate -acnv $NTPHOST the -a option uses the adjtime(2) call to gradually skew the local time to the remote time rather than just hopping. -- Best regards, Jörg-Volker. -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/jekmsk$ct6$1...@dough.gmane.org
Re: Message about ntp when updating
On Mi, 11 ian 12, 10:26:48, Bob Proulx wrote: For reasons that I haven't been following there has been a push to remove those shutdown links generally from everything. I don't know why. Perhaps someone else will comment on why. Removing those links seems like it breaks going to single user mode directly from multiuser mode but I only go to single user mode from a reboot anyway so I guess that is okay. I am sure the reasoning is that the reboot or halt will kill everything no matter what so might as well get to it as quickly as possible. On a server machine this wouldn't ever be an issue but waiting for a laptop to shutdown could be annoying. From lurking on debian-devel I can tell that your guess is right on :) (the motive is to speed-up the shutdown). Kind regards, Andrei -- Offtopic discussions among Debian users and developers: http://lists.alioth.debian.org/mailman/listinfo/d-community-offtopic signature.asc Description: Digital signature
Re: Message about ntp when updating
Andrei Popescu wrote: Bob Proulx wrote: For reasons that I haven't been following there has been a push to remove those shutdown links generally from everything. I don't know why. Perhaps someone else will comment on why. Removing those links seems like it breaks going to single user mode directly from multiuser mode but I only go to single user mode from a reboot anyway so I guess that is okay. I am sure the reasoning is that the reboot or halt will kill everything no matter what so might as well get to it as quickly as possible. On a server machine this wouldn't ever be an issue but waiting for a laptop to shutdown could be annoying. From lurking on debian-devel I can tell that your guess is right on :) (the motive is to speed-up the shutdown). Is there a plan for packages to deal with the links that are left behind? Could we, should we, start filing bugs if a package doesn't make the transition cleanly? Is there an explicit abandonment of going to single user mode directly from multuser mode? This certainly breaks it implicitly. However since I always reboot first anyway I can't really complain since I already didn't trust it for that purpose. Thanks, Bob signature.asc Description: Digital signature
Re: Message about ntp when updating
On Mi, 11 ian 12, 15:01:57, Bob Proulx wrote: From lurking on debian-devel I can tell that your guess is right on :) (the motive is to speed-up the shutdown). Is there a plan for packages to deal with the links that are left behind? Could we, should we, start filing bugs if a package doesn't make the transition cleanly? Is there an explicit abandonment of going to single user mode directly from multuser mode? This certainly breaks it implicitly. However since I always reboot first anyway I can't really complain since I already didn't trust it for that purpose. Err... no, ... seems sensible to me (unless there are technical reasons why cleanup is not possible)... not that I know of. But all this is from memory, you should probably ask the insserv maintainer or on -devel. Kind regards, Andrei -- Offtopic discussions among Debian users and developers: http://lists.alioth.debian.org/mailman/listinfo/d-community-offtopic signature.asc Description: Digital signature
Re: Message about ntp when updating
Jörg-Volker Peetz jvpe...@web.de wrote: Here I'm using rdate -an in a cron job with something like the following rdate -acnv $NTPHOST Just like ntpdate or any of the other cron based solutions proposed in this thread, this does not train the clock. So although rdate slews the clock, it doesn't train it to become more accurate over time. NTP does all this out of the box - and you get multiple server resiliency from the NTP Pool. (Do you really need the -c flag to rdate? My systems don't.) Chris -- 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/nkh1u8xnc5@news.roaima.co.uk
Re: Message about ntp when updating
On Jan 11, 2012, at 10:17 AM, Sthu Deus wrote: Good time of the day, Patrick. You worte: I see this message when running an update: Installing new version of config file /etc/cron.daily/ntp ... insserv: warning: current stop runlevel(s) (0 1 6) of script `ntp' overwrites defaults (empty). Should I be concerned about it? And, if so, what am I supposed to do to correct it. Not answering to Your question, but as mine opinion - if You do not intend other hosts to be sync-ing time w/ this host, I think You need the service at all. For sync-ing time from Internet for the host You can use ntpdate in cron/anacron for example. One more reason not to use ntpdate: It's no longer supported but the upstream developers of the ntp package. The ntp daemon has been enhanced so that it can now do all the things people used to use ntpdate for. Then to simplify maintenance (no longer have to maintain/ synchronize two code bases for doing the same things) ntpdate has been dropped from future versions of the ntp package. Rick -- 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/3554ebee-7d30-49de-9ac6-d0f85dfc6...@pobox.com
