Brief about lxd cluster master and slave

LXD
1

Hi everyone,
I have clustered with 3 nodes, can you explain how it works?
when nodeA stops which node becomes leader?
if nodeC stops which node becomes leader?
Are the database of container are stored in a separate node or each node has database?

A | https://192.168.1.34:8443 | YES | ONLINE | fully operational | x86_64 |
±-----------±---------------------------±---------±-------±------------------±-------------+
| B | https://192.168.1.244:8443 | YES | ONLINE | fully operational | x86_64 |
±-----------±---------------------------±---------±-------±------------------±-------------+
| B | https://192.168.1.192:8443 | YES | ONLINE | fully operational | x86_64 |
±-----------±---------------------------±---------±-------±------------------±-------------+

2

All 3 of your servers are marked as database servers so they all have a copy of the database, who the leader is at any one time is an internal detail which isn’t exposed and can change at any time.

3

ok. Does data installed in nodeA is transferred to nodeB, when nodeA is down?

4

No, all database servers have a full copy of the current database.

5

which means there is no data transfers between nodeA, B and C isit?

6

All 3 of them constantly communicate when in normal operation, if one goes down the remaining two continue to work and the one that went away will sync back the database transactions when it comes back online.

7

ok. For eg: NodeA has nginx and it goes down, what happens next? Does nodeB wil sync the nginx from nodeA?

8

No, instances that are on a server which has gone offline will remain unavailable until that server comes back up.

If storage is backed by Ceph, you can if needed move instances from an offline server to an online server and start them back up, but that’s only possible with Ceph storage as all other storage backends are local to the server and so unreachable when the server is offline.

9

ok thank you so much…!!!