THIS IS A TEST INSTANCE. ALL YOUR CHANGES WILL BE LOST!!!!
...
Note that the way broker handles PurgeRequest
is similar to how it handles ProduceRequest
with ack = -1 all and isr=all_live_replicas, e.g. the leader waits for all followers to catch up with its log_begin_offset
, doesn't expose message below log_begin_offset
, and checkpoints log_begin_offset
periodically. The low_watermark
of a partition will be the minimum log_begin_offset
of all replicas of this partition and this value will be returned to user in PurgeResponse
.
...
- If
log_begin_offset
of this partition on all live followers is larger than or equal to the offsetToPurge, the result of this partition will be itslow_watermark
, which is the minimumlog_begin_offset
of all its live replicas. - If
high_watermark
of this partition is smaller than the offsetToPurge, the result of this partition will beOffsetOutOfRangeException
.
If any replica of this partition goes offline, the result of this partition will beNotEnoughReplicasException
- If the leadership of this partition moves to another broker, the result of this partition will be
NotLeaderException
- If the result of this partition is not available after
RequestTimeoutMs
, the result of this partition will beTimeoutException
...