[ 
https://issues.apache.org/jira/browse/IGNITE-25645?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18012046#comment-18012046
 ] 

Alexander Lapin edited comment on IGNITE-25645 at 8/5/25 7:08 AM:
------------------------------------------------------------------

Following articles have been read
 * [https://dev.to/yugabyte/simulate-clock-skew-in-docker-container-2com]
 * 
[https://dev.to/yugabyte/crash-on-clock-skew-performance-vs-availability-5gp9]
 * 
[https://www.yugabyte.com/blog/accurate-clock-management-in-yugabytedb/#monitoring]

that showed the need to implement the following trivial algorithm:

If a node receives a message from the future with a time difference between the 
message and the local time on the node by more than MAX_CLOCK_SKEW, then this 
clearly indicates that the maximum allowed delta has been exceeded and thus we 
should print a warning message to the log.


was (Author: alapin):
Following articles have been read
 * [https://dev.to/yugabyte/simulate-clock-skew-in-docker-container-2com]
 * 
[https://dev.to/yugabyte/crash-on-clock-skew-performance-vs-availability-5gp9]
 * 
[https://www.yugabyte.com/blog/accurate-clock-management-in-yugabytedb/#monitoring]

that showed the need to implement the following trivial algorithm:

If a node receives a message from the future with a time difference between the 
message and the local time on the node by more than
MAX_CLOCK_SKEW, then this clearly indicates that the maximum allowed delta has 
been exceeded and thus we should print a warning message to the log.

> Design a mechanism for detecting when the maximum clock skew has been exceeded
> ------------------------------------------------------------------------------
>
>                 Key: IGNITE-25645
>                 URL: https://issues.apache.org/jira/browse/IGNITE-25645
>             Project: Ignite
>          Issue Type: Task
>            Reporter: Alexander Lapin
>            Assignee: Alexander Lapin
>            Priority: Major
>              Labels: ignite-3
>
> As an initial step of clock skew management we may log warn message and add a 
> metric on maximum clock skew exceedance. Within this ticket it's required to 
> define precise and effective mechanism for clock skew exceedance detection.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to