[Verse 1] Three nodes minimum, that's the magic number ZooKeeper ensemble won't split or slumber Odd counts prevent the dreaded brain split scenario Two gigs of RAM, four cores for the show CPU steady, memory consistent Persistent volumes make data resistant XFS or ext4, fast disks are key Network latency under ten millisecond spree [Chorus] Size it right, store it tight ZooKeeper needs resources burning bright Tick time, sync time, parameters aligned Keep your cluster's heartbeat refined Size it right, store it tight Configuration tuned for cluster flight Memory maps and journal logs combined ZooKeeper's secrets now defined [Verse 2] InitLimit sets the follower's first chance SyncLimit keeps the heartbeat dance TickTime controls the basic pulse Two thousand millis, that's the default result MaxClientConnections, sixty by design PreAllocSize sixty-four meg baseline SnapCount one hundred thousand transactions deep AutoPurge keeps old snapshots from heap [Chorus] Size it right, store it tight ZooKeeper needs resources burning bright Tick time, sync time, parameters aligned Keep your cluster's heartbeat refined Size it right, store it tight Configuration tuned for cluster flight Memory maps and journal logs combined ZooKeeper's secrets now defined [Bridge] ForceSync true for durability's sake JVM heap size, careful choices make Garbage collection tuned for low pause time Storage IOPS matter, performance climbs DataDir and DataLogDir split apart Transaction logs need their dedicated heart [Verse 3] MinSessionTimeout, MaxSessionTimeout range Forty seconds, four minutes, don't rearrange GlobalOutstandingLimit, one thousand max Client request queue, prevent overload attacks Strimzi operator handles the heavy lifting Kafka depends on ZooKeeper's shifting Persistent volume claims hold the state Configuration maps seal the ensemble's fate [Chorus] Size it right, store it tight ZooKeeper needs resources burning bright Tick time, sync time, parameters aligned Keep your cluster's heartbeat refined Size it right, store it tight Configuration tuned for cluster flight Memory maps and journal logs combined ZooKeeper's secrets now defined [Outro] Three nodes strong, parameters flowing ZooKeeper's wisdom, Kafka's growing Storage persistent, tuning complete Your distributed system can't be beat
← 1 Kafka Broker Configuration | 3 KRaft Mode (ZooKeeper-less Kafka) →