Cron running nine hours ahead, hwclock being set somehow? - Linux

This is a discussion on Cron running nine hours ahead, hwclock being set somehow? - Linux ; Greetings! Okay, I need assistance, even if it's just what to try next! :-) I have checked google, and I have found a few messages (with what I searched with) and it did give me a couple of ideas to ...

+ Reply to Thread
Results 1 to 10 of 10

Thread: Cron running nine hours ahead, hwclock being set somehow?

  1. Cron running nine hours ahead, hwclock being set somehow?


    Greetings!

    Okay, I need assistance, even if it's just what to try next! :-)

    I have checked google, and I have found a few messages (with what I
    searched with) and it did give me a couple of ideas to try, and I've tried
    them. I now have a couple of specific observations, and would like to
    know what should I try next?

    I'm running RH 7.1. Yesterday, I started running NTP. (I downloaded
    it and compiled it and started running it on my Linux system, and I
    already had it and am running it on the Sun/Solaris system.) The Linux
    system is running the NTP server, because I have an atomic clock monitor
    hooked up to the Linux system.

    On the Linux system, here is an ntpq -p:

    remote refid st t when poll reach delay offset jitter
    ================================================== ============================
    *LOCAL(0) LOCAL(0) 10 l 57 64 377 0.000 0.000 0.001

    On the Sun/Solaris system (the client), here is an ntpq -p:

    remote refid st t when poll reach delay offset disp
    ================================================== ============================
    barrycon LOCAL(0) 11 u 94 1024 375 0.61 9439.73 1512.51

    Now, also, just yesterday, I noticed I didn't have a timezone set.
    (!) (I did originally, and then I re-installed Linux, quite some time
    ago, and had never noticed!) :-( So, I set the TZ variable to
    "America/New_York" with the tzselect command in root.

    I then had to reset system time all over the place, but I understood
    this! :-)

    I have discovered, overnight/since yesterday, that cron is running
    nine hours ahead of time. Now, when I issue the command "date ; date -u"
    I get what I find believable and good, such as:

    Sat Sep 27 17:45:55 EDT 2003
    Sat Sep 27 21:45:55 UTC 2003

    (5:45:55 WAS the local time, and the UTC is four hours ahead, as it
    should be, when EST is on daylight savings time.)

    When I issued "hwclock --show" while ago, it was, sure enough, nine
    hours ahead. So, I have issued things like "hwclock --systohc" and
    "hwclock --hctosys" and now "hwclock --show" shows:

    Sat 27 Sep 2003 05:47:50 PM EDT 0.292312 seconds

    (And that IS the correct time when this was issued.)

    The man page for hwclock said a --hctosys would set the kernel's
    timezone value to the local timezone as indicated by the TZ environment
    variable, which WAS "America/New_York" when the command was run.

    I restarted cron, and saw no difference. I rebooted Linux -- even
    turning the system off and unplugging it for a few seconds (which I don't
    even normally ever power the system down) to just assure the hardware
    clock, which had just been set correctly, so far as I could tell, was what
    started the system time/clock.

    After I reboot Linux, the hwclock is back to nine hours ahead. Here
    is what a "hwclock --show" displayed, and the time was 18:02:56 at the
    time:

    Sun 28 Sep 2003 03:02:56 AM EDT 0.629749 seconds

    This is the time that cron is set to.

    When I shut down, I did see something that I figure probably lent
    itself to having my hardware clock get so far off -- in the two years
    since I installed this Linux! I saw this:

    Syncing hardware clock to system time modprobe: modprobe: Can't locate
    module char-major-10-135

    But, I can assure you that the hardware clock had been set to the
    system time, with hwclock --systohc, right beforehand. And, after doing
    this, a hwclock --show displayed the correct time.

    I've gone through the things I saw in google, but I'm out of ideas.
    Can anyone help me know why my hardware clock is being reset to nine hours
    ahead in the process of my shutting down and rebooting, even though it was
    set correct, in root, right beforehand?

    I've only specified what I felt was appropriate, in this message. If
    you need information I have failed to provide, please just nicely let me
    know, and I'll gladly provide it.

    Thank you!

    Barry
    --
    Barry L. Bond
    bbond@cfl.rr.com <- personal
    72235,1530 (CIS) <- personal

  2. Re: Cron running nine hours ahead, hwclock being set somehow?


    Greetings again!

    I'm adding something. I have the problem fixed, in a temporary
    fashion. When I rebooted Linux again, I pressed DEL and went into the
    BIOS and, sure enough, the time was wrong. I changed it here, and then
    rebooted, and the time is correct.

    This doesn't change the fact that I wouldn't mind knowing why my
    hwclock didn't do what I was expecting, so I still wouldn't mind hearing
    anything you may have to suggest on it. But, at least my cron is on the
    correct time now! :-) (sendmail also had the incorrect time, before.)

    Also, on the char-major-10-135 in my last message, I had noticed it
    before, but I had also forgotten it. I don't reboot my Linux system very
    often! :-) I went back in my kernel and did a 'make xconfig' and looked
    around, and I only saw one thing that dealt with RTC, and the answer is Y.
    If it IS supposed to be some module that is with my kernel, I'm not sure
    why it isn't present. (A locate char-major doesn't display anything.)

    That's the update! This isn't as critical now, but the main question
    I asked in the original posting still applies!

    Thank you!

    Barry
    --
    Barry L. Bond
    bbond@cfl.rr.com <- personal
    72235,1530 (CIS) <- personal

  3. Re: Cron running nine hours ahead, hwclock being set somehow?


    Greetings yet again!

    Well, my "inner/system" time is nine hours ahead again! (!) When I
    last rebooted, I pressed DEL and went into the BIOS, found the time wrong,
    set it correctly. I booted, and everything was good.

    Not too long afterward, though I can't say for sure exactly how long
    (but it wasn't an hour), the time was incorrect again. :-(

    The only thing that is done on my Linux system to change the time is
    the software I wrote to set the time according to the atomic clock
    monitor. However, this software has been running ever since I got the
    Linux system, late 1999, and I didn't experience this problem until after
    I set the timezone, on Saturday.

    The problem is more urgent again. I would like ANY suggestions as to
    what to check out or investigate to solve this problem.

    Thank you!

    Barry
    --
    Barry L. Bond
    bbond@cfl.rr.com <- personal
    72235,1530 (CIS) <- personal

  4. Re: Cron running nine hours ahead, hwclock being set somehow?

    This is kind of fascinating. Faced with weird stuff like this
    I usually start killing or disabling things until the problem stops.
    I've heard of daemon programs (possibly another ntp) getting launched
    that might not be very obvious. A ps -aux should show that. I've
    also heard of cron jobs getting launched that mess around with the
    clock.
    I had a problem with ntp that was more bother than problem,
    but seemed to point toward another ntp program running somehow.
    Upgraded to latest version and it went away. Of course, it is
    possible that a clean-up of some cron jobs might have solved that,
    too.

    On Mon, 29 Sep 2003 12:12:32 GMT, barry@barrycon.cfl.rr.com (Barry L.
    Bond) wrote:

    >
    >Greetings yet again!
    >
    > Well, my "inner/system" time is nine hours ahead again! (!) When I
    >last rebooted, I pressed DEL and went into the BIOS, found the time wrong,
    >set it correctly. I booted, and everything was good.
    >
    > Not too long afterward, though I can't say for sure exactly how long
    >(but it wasn't an hour), the time was incorrect again. :-(
    >
    > The only thing that is done on my Linux system to change the time is
    >the software I wrote to set the time according to the atomic clock
    >monitor. However, this software has been running ever since I got the
    >Linux system, late 1999, and I didn't experience this problem until after
    >I set the timezone, on Saturday.
    >
    > The problem is more urgent again. I would like ANY suggestions as to
    >what to check out or investigate to solve this problem.
    >
    > Thank you!
    >
    > Barry
    >--
    >Barry L. Bond
    >bbond@cfl.rr.com <- personal
    >72235,1530 (CIS) <- personal


    ------------
    Remove the underscores from the e-mail address.

  5. Re: Cron running nine hours ahead, hwclock being set somehow?

    "Barry L. Bond" wrote in message
    news:B2odb.27468$eS5.24372@twister.tampabay.rr.com ...
    >
    > Greetings!
    >
    > Okay, I need assistance, even if it's just what to try next! :-)
    >
    > I have checked google, and I have found a few messages (with what I
    > searched with) and it did give me a couple of ideas to try, and I've tried
    > them. I now have a couple of specific observations, and would like to
    > know what should I try next?
    >
    > I'm running RH 7.1. Yesterday, I started running NTP. (I downloaded
    > it and compiled it and started running it on my Linux system, and I
    > already had it and am running it on the Sun/Solaris system.) The Linux
    > system is running the NTP server, because I have an atomic clock monitor
    > hooked up to the Linux system.
    >
    > On the Linux system, here is an ntpq -p:
    >
    > remote refid st t when poll reach delay offset

    jitter
    >

    ================================================== ==========================
    ==
    > *LOCAL(0) LOCAL(0) 10 l 57 64 377 0.000 0.000

    0.001
    >
    > On the Sun/Solaris system (the client), here is an ntpq -p:
    >
    > remote refid st t when poll reach delay offset

    disp
    >

    ================================================== ==========================
    ==
    > barrycon LOCAL(0) 11 u 94 1024 375 0.61 9439.73

    1512.51
    >
    > Now, also, just yesterday, I noticed I didn't have a timezone set.
    > (!) (I did originally, and then I re-installed Linux, quite some time
    > ago, and had never noticed!) :-( So, I set the TZ variable to
    > "America/New_York" with the tzselect command in root.
    >
    > I then had to reset system time all over the place, but I understood
    > this! :-)
    >
    > I have discovered, overnight/since yesterday, that cron is running
    > nine hours ahead of time. Now, when I issue the command "date ; date -u"
    > I get what I find believable and good, such as:
    >
    > Sat Sep 27 17:45:55 EDT 2003
    > Sat Sep 27 21:45:55 UTC 2003
    >
    > (5:45:55 WAS the local time, and the UTC is four hours ahead, as it
    > should be, when EST is on daylight savings time.)
    >
    > When I issued "hwclock --show" while ago, it was, sure enough, nine
    > hours ahead. So, I have issued things like "hwclock --systohc" and
    > "hwclock --hctosys" and now "hwclock --show" shows:
    >
    > Sat 27 Sep 2003 05:47:50 PM EDT 0.292312 seconds
    >
    > (And that IS the correct time when this was issued.)
    >
    > The man page for hwclock said a --hctosys would set the kernel's
    > timezone value to the local timezone as indicated by the TZ environment
    > variable, which WAS "America/New_York" when the command was run.
    >
    > I restarted cron, and saw no difference. I rebooted Linux -- even
    > turning the system off and unplugging it for a few seconds (which I don't
    > even normally ever power the system down) to just assure the hardware
    > clock, which had just been set correctly, so far as I could tell, was what
    > started the system time/clock.
    >
    > After I reboot Linux, the hwclock is back to nine hours ahead. Here
    > is what a "hwclock --show" displayed, and the time was 18:02:56 at the
    > time:
    >
    > Sun 28 Sep 2003 03:02:56 AM EDT 0.629749 seconds
    >
    > This is the time that cron is set to.
    >
    > When I shut down, I did see something that I figure probably lent
    > itself to having my hardware clock get so far off -- in the two years
    > since I installed this Linux! I saw this:
    >
    > Syncing hardware clock to system time modprobe: modprobe: Can't locate
    > module char-major-10-135
    >
    > But, I can assure you that the hardware clock had been set to the
    > system time, with hwclock --systohc, right beforehand. And, after doing
    > this, a hwclock --show displayed the correct time.
    >
    > I've gone through the things I saw in google, but I'm out of ideas.
    > Can anyone help me know why my hardware clock is being reset to nine hours
    > ahead in the process of my shutting down and rebooting, even though it was
    > set correct, in root, right beforehand?
    >
    > I've only specified what I felt was appropriate, in this message. If
    > you need information I have failed to provide, please just nicely let me
    > know, and I'll gladly provide it.
    >
    > Thank you!
    >
    > Barry
    > --
    > Barry L. Bond
    > bbond@cfl.rr.com <- personal
    > 72235,1530 (CIS) <- personal

    Is by any chance America/New_York 9 hours out from GMT ? Maybe that is where
    your problem is. Some Linux programs assume your clock is set to GMT &
    calculates the local time. Some OSs assume its local time. Most of the Linux
    programs I have looked at have some setting to let you use the clock as
    local time & not calculate. I think when the pc runs the hw clock is ignored
    & the kernel keeps time via regular interrupts. When you power down the pc
    the hw clock is updated though this may happen at other possibly scheduled
    intervals. Try setting the time using whatever command it is (cant remember
    & I work in an AS/400 & Windows place). That should set the hw clock to be 9
    hours ahead of local time & local time to be true or hw clock to be the same
    depending on the options you take. Make sure this matches whatever the RH
    setup thinks you have. Try scanning /etc/rc.d for hwclock & its friends.

    --
    Jonathan.



  6. Re: Cron running nine hours ahead, hwclock being set somehow?

    In article <3f78c45f.768459776@news-server.insight.rr.com>,
    Joe Fischer wrote:

    > This is kind of fascinating. Faced with weird stuff like this
    >I usually start killing or disabling things until the problem stops.
    >I've heard of daemon programs (possibly another ntp) getting launched
    >that might not be very obvious. A ps -aux should show that. I've
    >also heard of cron jobs getting launched that mess around with the
    >clock.


    Hi Joe!

    Thank you for trying to help me!

    There appeared to be only one ntp (ntpd) running. I just killed it
    on the Linux. Since I did make two changes, both on Saturday, and it's
    from then on that I've had this problem with cron and sendmail running
    nine hours ahead, I'm trying for up to about a day without ntpd running on
    Linux.

    Now, I also set the timezone on Saturday. And, it wasn't set at all,
    and I prefer it to be (I had it set in my last installation). So, now I'm
    running with the timezone set to "America/New_York" -- set by the tzselect
    command in root, and, temporarily, no ntp.

    Here's what I find at the moment, in my experimentation...

    The hwclock is actually NOT set forward! It's just cron and
    sendmail, but it does NOT appear to be the system's time!

    Here are a few commands, taken directly from my root window:

    [root@barrycon /root]# hwclock --local
    Tue 30 Sep 2003 03:35:59 PM EDT 0.354882 seconds
    [root@barrycon /root]# hwclock --utc
    Tue 30 Sep 2003 11:36:01 AM EDT 0.667217 seconds
    [root@barrycon /root]# date ; date -u
    Tue Sep 30 15:35:26 EDT 2003
    Tue Sep 30 19:35:26 UTC 2003

    The hwclock --local displayed the correct local time. It IS not
    incorrect! (At least, not horribly off. It's already off a few seconds,
    but certainly not nine hours!) :-O :-)

    The hwclock --utc is... "interesting." The eastern time zone is four
    hours BEHIND GMT when we're in daylight savings time, as we are now until
    the last Sunday of October. (It's FIVE hours behind GMT when we're in
    standard, as opposed to daylight savings, time.) But, it shows UTC as
    four hours earlier than GMT!

    Yet, you'll see the next command "date ; date -u" I issue. The date
    command shows the time in my time zone (and it's EDT -- correct) and the
    date -u shows the UTC time. THIS IS 100% CORRECT! It shows my local time
    as EDT (Eastern Daylight Time), and it shows it four hours BEHIND UTC, as
    it should be.

    The hwclock command showed EDT in both cases, even though it showed
    four hours (but EARLIER instead of LATER) for GMT/UTC. I'll read through
    man pages (again, I certainly have been!) and see if I can see some time
    zone setting that is different for hwclock than the rest of the system
    time/date commands! :-/

    Does this prompt any new ideas from you?

    > I had a problem with ntp that was more bother than problem,
    >but seemed to point toward another ntp program running somehow.
    >Upgraded to latest version and it went away. Of course, it is
    >possible that a clean-up of some cron jobs might have solved that,
    >too.


    At home, there are only two people who use my systems, and it's just
    me and my mother. And, while there are a few cron jobs for my mother, I
    have set them up. She doesn't do that! :-) So, I am extremely familiar
    with all my cron jobs. I set up all of them! :-) All the jobs that run
    in barry are basically just reminders for placing certain things in my
    Financial Management System software that are paid the same date each
    month, etc. Just basic shell scripts of commands. All that run in root
    except for one (GC1000_Update) just run things that should be run from
    root, such as an automatic backup (using BRU) every Saturday morning,
    going to my weather station a few minutes after midnight each day to get
    "yesterday's" record weather data for storing in a file, etc. (Which, by
    the way, makes the time on the weather station nine hours ahead, because
    the program sets the weather station's time after it reads a complete
    record, because the weather station loses time significantly as it prints
    a complete record to the serial port.) The GC1000_Update is a daemon that
    I wrote myself, which runs at 5:02 and 17:02 (5:02 AM and PM, or twice a
    day) and sets the Linux system time to the atomic clock monitor, which is
    connected to a serial port.

    There have been no changes in crontabs, and I've been running the
    GC1000 since 1999 with not a problem. From what I am thinking of or
    realizing at the moment, I just don't see how any of these are a problem.

    Thank you again for your help, Joe! If you have any other ideas, I'm
    open to them! :-)

    Barry
    --
    Barry L. Bond
    bbond@cfl.rr.com <- personal
    72235,1530 (CIS) <- personal

  7. Re: Cron running nine hours ahead, hwclock being set somehow?

    In article <3f795a69$0$10778$afc38c87@news.easynet.co.uk>, wrote:

    Hi Jonathan!

    I very much appreciate your trying to help me. Two (or more) heads
    are better than one! :-)

    >Is by any chance America/New_York 9 hours out from GMT ? Maybe that is where
    >your problem is. Some Linux programs assume your clock is set to GMT &
    >calculates the local time. Some OSs assume its local time. Most of the Linux
    >programs I have looked at have some setting to let you use the clock as
    >local time & not calculate. I think when the pc runs the hw clock is ignored
    >& the kernel keeps time via regular interrupts. When you power down the pc
    >the hw clock is updated though this may happen at other possibly scheduled
    >intervals. Try setting the time using whatever command it is (cant remember
    >& I work in an AS/400 & Windows place). That should set the hw clock to be 9
    >hours ahead of local time & local time to be true or hw clock to be the same
    >depending on the options you take. Make sure this matches whatever the RH
    >setup thinks you have. Try scanning /etc/rc.d for hwclock & its friends.


    No. The "America/New_York" time zone, or the eastern time zone, is
    four hours BEHIND GMT this time of the year, when we're in daylight
    savings time. It is FIVE hours behind GMT after the last Sunday in
    October, when we return to standard time. (GMT doesn't change at daylight
    savings time, but Eastern time changes from EST [Eastern Standard Time] to
    EDT [Eastern Daylight Time].)

    According to what I've seen in the man pages, it is highly
    recommended to have the timezone set, as I do now. (I set it Saturday.)
    I had it set in my last installation (RH 6.0) and all was well.

    I have this in another posting, but here is something I ran for Joe
    Fischer, who is also trying to help me:

    [root@barrycon /root]# hwclock --local
    Tue 30 Sep 2003 03:35:59 PM EDT 0.354882 seconds
    [root@barrycon /root]# hwclock --utc
    Tue 30 Sep 2003 11:36:01 AM EDT 0.667217 seconds
    [root@barrycon /root]# date ; date -u
    Tue Sep 30 15:35:26 EDT 2003
    Tue Sep 30 19:35:26 UTC 2003

    The hardware clock is NOT nine hours ahead! It's off a few seconds,
    but not nine hours! (That was, at least, very close to the time when I
    ran it.) But, the --local showed the correct local time. The --utc
    showed four hours, but UTC/GMT being four hours behind EDT, instead of the
    actual case of it being the other way around!

    The next command I executed, "date ; date -u" showed the current
    local date/time and then the UTC date/time. Both of these are 100%
    correct. EDT (Eastern Daylight Time) and UTC, being four hours ahead.

    So, it appears it's NOT the hardware clock. It appears it's NOT the
    system time (what date/date -u shows)! At the moment, I'm not sure what
    time cron and sendmail are operating on, but it's nine hours ahead! :-O

    Thank you for your assistance! I'm open to anything you or anyone
    else thinks of! :-)

    Barry
    --
    Barry L. Bond
    bbond@cfl.rr.com <- personal
    72235,1530 (CIS) <- personal

  8. Re: Cron running nine hours ahead, hwclock being set somehow?

    On Tue, 30 Sep 2003 20:24:24 GMT, barry@barrycon.cfl.rr.com (Barry L.
    Bond) wrote:

    >In article <3f795a69$0$10778$afc38c87@news.easynet.co.uk>, wrote:
    >
    >Hi Jonathan!
    >
    > I very much appreciate your trying to help me. Two (or more) heads
    >are better than one! :-)
    >
    >>Is by any chance America/New_York 9 hours out from GMT ? Maybe that is where
    >>your problem is. Some Linux programs assume your clock is set to GMT &
    >>calculates the local time. Some OSs assume its local time. Most of the Linux
    >>programs I have looked at have some setting to let you use the clock as
    >>local time & not calculate. I think when the pc runs the hw clock is ignored
    >>& the kernel keeps time via regular interrupts. When you power down the pc
    >>the hw clock is updated though this may happen at other possibly scheduled
    >>intervals. Try setting the time using whatever command it is (cant remember
    >>& I work in an AS/400 & Windows place). That should set the hw clock to be 9
    >>hours ahead of local time & local time to be true or hw clock to be the same
    >>depending on the options you take. Make sure this matches whatever the RH
    >>setup thinks you have. Try scanning /etc/rc.d for hwclock & its friends.

    >
    > No. The "America/New_York" time zone, or the eastern time zone, is
    >four hours BEHIND GMT this time of the year, when we're in daylight
    >savings time. It is FIVE hours behind GMT after the last Sunday in
    >October, when we return to standard time. (GMT doesn't change at daylight
    >savings time, but Eastern time changes from EST [Eastern Standard Time] to
    >EDT [Eastern Daylight Time].)
    >
    > According to what I've seen in the man pages, it is highly
    >recommended to have the timezone set, as I do now. (I set it Saturday.)
    >I had it set in my last installation (RH 6.0) and all was well.
    >
    > I have this in another posting, but here is something I ran for Joe
    >Fischer, who is also trying to help me:
    >
    >[root@barrycon /root]# hwclock --local
    >Tue 30 Sep 2003 03:35:59 PM EDT 0.354882 seconds
    >[root@barrycon /root]# hwclock --utc
    >Tue 30 Sep 2003 11:36:01 AM EDT 0.667217 seconds
    >[root@barrycon /root]# date ; date -u
    >Tue Sep 30 15:35:26 EDT 2003
    >Tue Sep 30 19:35:26 UTC 2003
    >
    > The hardware clock is NOT nine hours ahead! It's off a few seconds,
    >but not nine hours! (That was, at least, very close to the time when I
    >ran it.) But, the --local showed the correct local time. The --utc
    >showed four hours, but UTC/GMT being four hours behind EDT, instead of the
    >actual case of it being the other way around!
    >
    > The next command I executed, "date ; date -u" showed the current
    >local date/time and then the UTC date/time. Both of these are 100%
    >correct. EDT (Eastern Daylight Time) and UTC, being four hours ahead.
    >
    > So, it appears it's NOT the hardware clock. It appears it's NOT the
    >system time (what date/date -u shows)! At the moment, I'm not sure what
    >time cron and sendmail are operating on, but it's nine hours ahead! :-O
    >
    > Thank you for your assistance! I'm open to anything you or anyone
    >else thinks of! :-)
    >
    > Barry
    >--
    >Barry L. Bond
    >bbond@cfl.rr.com <- personal
    >72235,1530 (CIS) <- personal


    That reminded me that the kernel generally ignores the
    hardware clock after booting up. There might be some useful reading
    in rtc.txt in the Documentation folder of your kernel distro. There
    was some interesting debugging techniques at
    http://www.eecis.udel.edu/~mills/ntp/html/debug.html.
    I just have to wonder if ntp is doing this or something else
    is responsible.
    ------------
    Remove the underscores from the e-mail address.

  9. Re: Cron running nine hours ahead, hwclock being set somehow?

    Hi Joe!

    > That reminded me that the kernel generally ignores the
    >hardware clock after booting up. There might be some useful reading
    >in rtc.txt in the Documentation folder of your kernel distro. There
    >was some interesting debugging techniques at
    >http://www.eecis.udel.edu/~mills/ntp/html/debug.html.
    > I just have to wonder if ntp is doing this or something else
    >is responsible.


    Thank you for offering further suggestions! I really appreciate it!

    I did see the rtc.txt file. I will look over that, and I'll try to
    assure that the software I wrote years ago which sets the system time from
    the atomic clock monitor is good. I still don't think this is causing the
    current problem, though. I can go into the BIOS clock, set it properly,
    and boot, and for just a couple of minutes, all time looks good. Cron
    even starts at the correct time! But, within just a few minutes, it's
    nine hours ahead again! :-O And, no NTP is running. And my software to
    update the system time from the atomic clock monitor hasn't run.

    I see the NTP debugging info via the URL you provided. Thank you for
    that, too! I will look into it. However, at the moment, I have NTP
    stopped on both the Linux and Sun systems. It is not NTP. date and date
    -u show the exact proper time.

    However, when I get NTP started again, I'll be watching the time
    carefully for a little while, and it will be good to have access to
    debugging info such as you provided above.

    I haven't had much time the past couple of days to give to this, and
    I may not have much time before this upcoming weekend or even early next
    week. :-(

    I guess I'll look into the Timezone. That's the only other thing I
    changed, as of Saturday, and that is still in effect. I've had this cron
    and sendmail running nine hours ahead since I made the two changes to my
    Linux system Saturday.

    I'll spend time on this as soon as I can, and I'll let you know what
    I discover!

    Barry
    --
    Barry L. Bond
    bbond@cfl.rr.com <- personal
    72235,1530 (CIS) <- personal

  10. Re: Cron running nine hours ahead, hwclock being set somehow?

    Greetings!

    Well, I *think* my problem is fixed...

    It appears, though I *had* an /etc/localtime/it was set to something,
    when I read what was in the file, it looked correct to me, but it was
    considerably less than the file /usr/share/zoneinfo/EST5EDT, which I
    symbolically linked /etc/localtime to.

    After I performed this link, I rebooted the Linux system. (Again!
    I typically have gone many months between reboots, and this
    is like the third or fourth reboot in the past week!) :-)

    At the moment, sendmail and cron appear to be on the correct time.

    Now, again/reminder, when I executed date ; date -u before, I saw the
    100% correct EDT time and the 100% correct UTC time. However, the
    hardware clock showed the correct localtime if I executed hwclock --local,
    and a time four hours BEHIND local time (instead of later/after local
    time) if I executed hwclock --utc. So, I set my hardware clock to UTC,
    as it had been set to local before. Perhaps this made a difference, as
    well.

    So, two changes, my hardware clock is now set to UTC and
    /etc/localtime is linked to /usr/share/zoneinfo/EST5EDT.

    At the moment, I have NOT restarted the NTP (on the Sun or on the
    Linux) or executed my software which sets the Linux clock to the atomic
    clock, via the atomic clock monitor I have at my house with a serial
    connection. I'm still watching things. If things appear to be okay for
    another hour or so, I'll then execute my atomic clock monitor software,
    and check things, and I'll probably wait another day or so before I start
    running NTP again. (These are now going to be ONE at a time, as I look
    over things, so if there is a problem, I'll know which caused it!) :-)

    Thank you, everyone who has tried to help me in the past week!
    Unless I encounter a problem with the cron/sendmail or any other time
    after I execute my software or restart NTP, we can consider this matter
    closed! :-)

    Barry
    --
    Barry L. Bond
    bbond@cfl.rr.com <- personal
    72235,1530 (CIS) <- personal

+ Reply to Thread