Hi David, I’ll post my StackOverflow response here as well:
There is typically no need to adjust the member timeouts. I would, instead, suggest that you look for the root cause as to why your members are timing out. Since it sounds like they are good for a couple of hours before the issues arise, my first instinct would be to look at memory utilization and, specifically, GC pauses. Long GC pauses can easily result in member timeouts. --Jens From: David O'Loughlin - Yieldbroker <David.O'lough...@yieldbroker.com> Date: Monday, September 4, 2023 at 6:55 PM To: user@geode.apache.org <user@geode.apache.org> Subject: Disable network partition detection for two node cluster !! External Email Dear all, We use a 4 node cluster consisting of 2 cache servers and 2 locators in our production and test environments. But we also use a 2 node cluster consisting of a single cache server and locator in our development environments. We have constant problems with the cache server and/or the locator suiciding in 2 node cluster environments because of missed heartbeats. We don't have this problem in 4 node cluster environments. I would like some advice as to the wisdom of setting enable-network-partition-detection=false in our 2 node cluster environments to prevent accidental suicide. Many thanks, David !! External Email: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender.
