[
https://issues.apache.org/jira/browse/SLING-5326?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Stefan Egli updated SLING-5326:
-------------------------------
Description:
OAK-2682 introduces time-difference detection to the documentMk via a JMX
method {{determineServerTimeDifferenceMillis}} that determines how much the
local VM's clock differs from the VM of the DocumentStore.
There should be a Sling Health-Check that verifies that this
{{determineServerTimeDifferenceMillis}} is within reasonable limits.
Suggested defaults:
* low-water mark of 1 sec: when lower, fine, when higher, a warn is issued
* high-water mark of 5 sec: when higher, a critical message is issued
While the above is only verifying clocks in a documentMk cluster, the clocks
within the cross-cluster topology should also be verified. This can be achieved
independent of the underlying oak NodeStore:
* use (still existing) "{{created}}" json property of discovery's topology
announcement to check inter-cluster clocks. This can be checked in both
direction of the HTTP PUT. It should use higher water marks, since this
involved HTTP round-trips as well as it doesn't have the same criticality as
intra-cluster clocks. Suggested defaults here:
** low-water mark of 5 sec
** high-water mark of 10 sec
was:
OAK-2682 introduces time-difference detection to the documentMk via a JMX
method {{determineServerTimeDifferenceMillis}} that determines how much the
local VM's clock differs from the VM of the DocumentStore.
There should be a Sling Health-Check that verifies that this
{{determineServerTimeDifferenceMillis}} is within reasonable limits.
Suggested defaults:
* low-water mark of 1 sec: when lower, fine, when higher, a warn is issued
* high-water mark of 5 sec: when higher, a critical message is issued
While the above is only verifying clocks in a documentMk cluster, there is also
> introduce a time-difference health-check to discovery.oak
> ---------------------------------------------------------
>
> Key: SLING-5326
> URL: https://issues.apache.org/jira/browse/SLING-5326
> Project: Sling
> Issue Type: New Feature
> Components: Extensions
> Affects Versions: Discovery Oak 1.1.0
> Reporter: Stefan Egli
> Assignee: Stefan Egli
> Fix For: Discovery Oak 1.1.2
>
>
> OAK-2682 introduces time-difference detection to the documentMk via a JMX
> method {{determineServerTimeDifferenceMillis}} that determines how much the
> local VM's clock differs from the VM of the DocumentStore.
> There should be a Sling Health-Check that verifies that this
> {{determineServerTimeDifferenceMillis}} is within reasonable limits.
> Suggested defaults:
> * low-water mark of 1 sec: when lower, fine, when higher, a warn is issued
> * high-water mark of 5 sec: when higher, a critical message is issued
> While the above is only verifying clocks in a documentMk cluster, the clocks
> within the cross-cluster topology should also be verified. This can be
> achieved independent of the underlying oak NodeStore:
> * use (still existing) "{{created}}" json property of discovery's topology
> announcement to check inter-cluster clocks. This can be checked in both
> direction of the HTTP PUT. It should use higher water marks, since this
> involved HTTP round-trips as well as it doesn't have the same criticality as
> intra-cluster clocks. Suggested defaults here:
> ** low-water mark of 5 sec
> ** high-water mark of 10 sec
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
