# 3 node percona cluster - 2 nodes got out of sync but 1 node was still up

**URL:** <https://forums.percona.com/t/3-node-percona-cluster-2-nodes-got-out-of-sync-but-1-node-was-still-up/17605>\
**Category:** Percona XtraDB Cluster 8.x\
**Created:** [September 21, 2022, 6:03pm UTC](https://forums.percona.com/t/3-node-percona-cluster-2-nodes-got-out-of-sync-but-1-node-was-still-up/17605 "2022-09-21T18:03:15Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![jasonfe33](https://avatars.discourse-cdn.com/v4/letter/j/838e76/32.png) [@jasonfe33](https://forums.percona.com/u/jasonfe33)\
**Post date:** [September 21, 2022, 6:03pm UTC](https://forums.percona.com/t/3-node-percona-cluster-2-nodes-got-out-of-sync-but-1-node-was-still-up/17605/1 "2022-09-21T18:03:15Z")

</div>

We were under the impression that if 1 node remained in the cluster that it would shutdown because only 1 node was left. You would need to bootstrap the last node alive to make sure the other 2 nodes synced with it for most recent data. Is this still the case for clusters?

Issue we experience was had 2 nodes go out of sync (but mysql was still running) and after restarting mysql on the other 2 nodes all was good again. the only node that was left was saying it was primary. site however was having issues at that time but after the mysql restarts on 2 of the nodes all was good again.

for the gcomm://nodename,nodename,nodename can these be IP’s and or Hostnames?? or must they be IP’s? Does one way work better than the other and if so why?

---

<div class="post-metadata">

**Author:** ![Ivan\_Groenewold](https://sea1.discourse-cdn.com/flex019/user_avatar/forums.percona.com/ivan_groenewold/32/6299_2.png) [@Ivan\_Groenewold](https://forums.percona.com/u/Ivan_Groenewold)\
**Post date:** [September 23, 2022, 11:10am UTC](https://forums.percona.com/t/3-node-percona-cluster-2-nodes-got-out-of-sync-but-1-node-was-still-up/17605/2 "2022-09-23T11:10:24Z")

</div>

Hello, by design if nodes leave the cluster gracefully (which seems to be what happened as you did a normal shutdown of mysql) then the quorum is not lost, so even with 1 node the cluster will keep running in this case. For gcomm either IP or hostnames will work. You might want to use IPs to avoid relying on DNS service.

---

<div class="post-metadata">

**Author:** ![jasonfe33](https://avatars.discourse-cdn.com/v4/letter/j/838e76/32.png) [@jasonfe33](https://forums.percona.com/u/jasonfe33)\
**Post date:** [September 26, 2022, 9:47pm UTC](https://forums.percona.com/t/3-node-percona-cluster-2-nodes-got-out-of-sync-but-1-node-was-still-up/17605/3 "2022-09-26T21:47:41Z")

</div>

So, - two exited the cluster, _somehow_ (we don’t know why this happens, and thus far it is more regular than can be considered acceptable)

- we don’t know why the 1 primary node still reported OK given it shouldn’t have had a quorum (as we believe that it was _not_ due to a graceful exiting of the cluster).