0

we test scalein in enviorment with 3 kafka nodes each one has a broker and zookeeper service on in the replica.factor=2 after the service went down on 2 nodes the leader of the partition stayed the old one and didnt movve to the only node that is up

and a a error message is sent

Error: partition 0 does not have a leader. Skip getting offsets Error: partition 2 does not have a leader. Skip getting offsets Error: partition 3 does not have a leader. Skip getting offsets

why this could happen?

here you can see how the leaders stayed even though there is replica set

NoamiA
  • 521
  • 4
  • 19
  • As far as I understand Kafka can not find a new leader since the replica is 2, but there is only one node. – Ehud Lev Nov 05 '19 at 16:53

1 Answers1

0

According to kafka configuration

The replication factor for the offsets topic (set higher to ensure availability). Internal topic creation will fail until the cluster size meets this replication factor requirement.

I guess that kafka cluster can not serve or do anything with a topic that has replica factor that is bigger than it's cluster size

Ehud Lev
  • 2,461
  • 26
  • 38
  • but if I want to preform scale in its impossible with replica factor=2? – NoamiA Nov 07 '19 at 09:28
  • What do you mean by scale -> more parallelize / more data / more replication ? Replication factor is just a configuration for how many copies of your data you want – Ehud Lev Nov 07 '19 at 09:47
  • ok and if 2 nodes are down if I will truncate the data from topic and restart zookeeper and kafka service it sould worked? – NoamiA Nov 10 '19 at 10:55