1

I have already gone through the one of the post however, the recommendation was to run things in same LAN segment or may be swarm for docker Docker Geode remote locator

My question is if I run 2 locator on 2 different node (VMs or Physical hardware) and also specify --locators='hostN[port]' referencing one another, how should I know if they did form a cluster?

Do i need to configure them always in a WAN configurations, even if they are part of same LAN but independent node (and not processes within one node)?

list members is not showing both the locators.

I am able to connect from one gfsh console to the locator in a different node gfsh console. but as long as I am not giving any bind addresses while starting the locators, both my locator start however, I don;t know if they formed a cluster (or if they are connected to each other).

I am evaluating with Apache Geode but need HA across VM and not processes within the same node. Please guide. Thanks in advance.

marc_s
  • 732,580
  • 175
  • 1,330
  • 1,459
Dheeraj
  • 23
  • 6
  • Should I expect locators running on 2 different VMs starting up properly with corresponding --locators='vm1[port],vm2[port]' arguments should show up in list members? I was expecting them to be shown connected somewhere. Could someone please clarify? – Dheeraj Mar 19 '19 at 15:00

1 Answers1

1

I changed the IPv4 configuration and provided a nearby LAN address, recommendation is to bring them in the same LAN segment. If I start the locator virtually simultaneously and not wait for another locator to get up already, it seems to find it.

Other options I found was to increase: ack-wait-threshold=15 member-timeout=5000

But keep in mind, the more time you put in here for acknowledgement, the same time it will take later to tell you the node is not responding in case it fails for any odd reason.

Dheeraj
  • 23
  • 6