Stefan Egli created SLING-5290:
----------------------------------
Summary: Ensure heartbeat self-check works with any
timeout/interval config
Key: SLING-5290
URL: https://issues.apache.org/jira/browse/SLING-5290
Project: Sling
Issue Type: Improvement
Components: Extensions
Affects Versions: Discovery Impl 1.2.0
Reporter: Stefan Egli
Assignee: Stefan Egli
Priority: Minor
Fix For: Discovery Impl 1.2.2
SLING-5195 introduced a 'heartbeat self-check': a separate thread that checks
if the local instance is successfully storing heartbeats to the repository.
That check fails if the time since last heartbeat is 'too long'.
SLING-5284 fiddled with what 'too long' exactly means: namely that it should
take into account not only the timeout but also the interval, as a result
{{timeout - 2*interval}} was used.
Although not very likely, but lets say you use {{timeout=60sec}},
{{interval=25sec}}. That would mean the self-check fails after 10sec already -
which is clearly too quickly. This formula needs to be improved. And actually
tests (ClusterTest.testDuplicateInstance3726) highlighted this exact problem.
A new formula should be:
* at minimum {{2*interval}} (or {{timeout}} if that's even lower)
* at maximum {{timeout - 2*interval}}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
