This discussion is archived
8 Replies Latest reply: Jan 17, 2013 7:16 AM by Levi-Pereira RSS

Time difference between nodes extreme different ;  how to correct it ?

Bernd Niewels Newbie
Currently Being Moderated
Following Situation:

Example: Time 'now' 2013/01/15 03:00 PM

Node 1 has Time : 2013/01/15 02:15 PM (45 min 'behind' time now)

Node 2 has Time: 2013/01/16 04:05 AM (14 h 'before' time now)


Cluster and Database works fine (Client works well, Application for Database works fine).


What are 'Best practise' to correct the time ?
  • 1. Re: Time difference between nodes extreme different ;  how to correct it ?
    N88 Newbie
    Currently Being Moderated
    Hi there,




    I hope this link is helpful in answering your question.
    http://www.reachdba.com/showthread.php?388-Time-Difference-between-RAC-Nodes


    Please let me know if you have any further question.


    -Thanks,
    -N

    Edited by: N88 on Jan 15, 2013 9:20 AM
  • 2. Re: Time difference between nodes extreme different ;  how to correct it ?
    rcc50886 Journeyer
    Currently Being Moderated
    What is the version of cluster you using ?

    using NTP on all nodes is the best practice to overcome any issues related to cluster in this case CTSS will start in observer mode. If NTP is not running then CTSS process will start and it maintenance time synchronizing between nodes.

    even with CTSS if time difference between nodes exceeds the certain limit then node will be ejected from the cluster.
  • 3. Re: Time difference between nodes extreme different ;  how to correct it ?
    Levi-Pereira Guru
    Currently Being Moderated
    Hi,

    If downtime is a option then is recommend stop/fix/start cluster. (less than 3 min of downtime)

    Stop Cluster fix time on all nodes and Start Cluster.


    Please read:
    RAC: Frequently Asked Questions [ID 220970.1]
    * Does Oracle RAC work with NTP (Network Time Protocol)?
    Keep the following points in mind:
    .. . If you intend on making large time changes it is best to shutdown the instances and the entire Oracle Clusterware stack on that node to avoid a false eviction, especially if you are using the Oracle RAC 10g low-brownout patches, which allow really low misscount settings ...

    Regards,
    Levi Pereira
  • 4. Re: Time difference between nodes extreme different ;  how to correct it ?
    Bernd Niewels Newbie
    Currently Being Moderated
    At this moment i think of this:

    First node (45min 'behind' actual time) shutdown, correct the BIOS-Time .. start the node.
    Make Shure ntp works fine (at now i have no chance, because ntpq quits while only accept 600s difference ..).

    If First Node running 'well' then

    Second Node: Shutdown (wait 15 hours, because now 14 hour 'before' actual time), correct BIOS-Time , start node.
    Make Shure ntp works fine (at now i have no chance, because ntpq quits while only accept 600s difference ..).

    I am not shure, if that 'works'.
  • 5. Re: Time difference between nodes extreme different ;  how to correct it ?
    Bernd Niewels Newbie
    Currently Being Moderated
    at this time the only chance to be shure ntpd is running .. is to configure only local clock. Not NTP-Clock over ethernet.

    NTP is Running with -x -u option. Local Clock is used .. and Oracle runs on both nodes.
    No Imact with that time difference since now .. ( i also have 'questionmarks' in my brain ..)

    System is: 11G2 11.0.2.3.0

    Perhaps my scenario works (see post above).

    I dont see any chance to do it in another way right now. Perhaps someone knows a little bit more ..
  • 6. Re: Time difference between nodes extreme different ;  how to correct it ?
    Levi-Pereira Guru
    Currently Being Moderated
    Bernd Niewels wrote:
    At this moment i think of this:

    First node (45min 'behind' actual time) shutdown, correct the BIOS-Time .. start the node.
    Make Shure ntp works fine (at now i have no chance, because ntpq quits while only accept 600s difference ..).

    If First Node running 'well' then

    Second Node: Shutdown (wait 15 hours, because now 14 hour 'before' actual time), correct BIOS-Time , start node.
    Make Shure ntp works fine (at now i have no chance, because ntpq quits while only accept 600s difference ..).

    I am not shure, if that 'works'.
    The procedure is correct.
    I mean ( Stop whole cluster, fix time on first node and start cluster on that node, fix time on second node and start cluster).
    I see no problem in start second node after "stop cluster/adjust time". You don't need wait 15 min to start it the only issue you may face is a time "jump" in the logfiles, making it more difficult to analyze them.

    You should check issue about adjust time on your application (Maybe some table which depends on correct time to work), if yes you must wait 15 min until start database on both nodes. I believe this is not your case because you have big diff on time and everything looks like well on application.

    P.S :
    (wait 15 hours, because now 14 hour 'before' actual time)
    You mean minutes, right?

    Regards,
    Levi Pereira

    Edited by: Levi Pereira on Jan 16, 2013 10:36 AM
  • 7. Re: Time difference between nodes extreme different ;  how to correct it ?
    Bernd Niewels Newbie
    Currently Being Moderated
    realy 'hours'. (see example ... it is real ..)
    And timezone is the same.

    I only the guy who found out (installed from 'another' .. and system be well since 1 1/2 years).

    No i have to correct it. Without damage of cluster and database.
    I am not worry about the node in the 'past time' ... i am worry about the node which are in 'future time'.
  • 8. Re: Time difference between nodes extreme different ;  how to correct it ?
    Levi-Pereira Guru
    Currently Being Moderated
    Bernd Niewels wrote:
    realy 'hours'. (see example ... it is real ..)
    And timezone is the same.

    I only the guy who found out (installed from 'another' .. and system be well since 1 1/2 years).

    No i have to correct it. Without damage of cluster and database.
    I am not worry about the node in the 'past time' ... i am worry about the node which are in 'future time'.
    Far as I know, you can foward or backward time without worry about clusterware stack (during time change the cluster should be down).
    About database it's a lilttle more complex, you must be aware of this note:
    *What are the effects of changing the system clock on an Oracle Server instance? [ID 77370.1]*
    Oracle RDBMS server is immuned to time changes, i.e. will not affect transaction/read consistency operations.

    If you have doubts on the impact of this procedure, I recommend opening a SR on MoS.

    Regards,
    Levi Pereira

Legend

  • Correct Answers - 10 points
  • Helpful Answers - 5 points