retry delivering msg to the client, regardless of the reset result
fix channel creation returns 200 when the topic is not found
v0.3.7-HA.1.12.6
3 years ago
fix load channel should not persist the init meta which will cause the state lost for the channel if exit by unexpected.
handle the message which timeout too much to avoid blocking too long
add the experimental kv-index for the topic (disabled by default).
clean check max available if can not greedy clean directly
return an error if update channel state or delete channel failed to sync
v0.3.7-HA.1.12.5
3 years ago
fix: etcd watched max index which may cause lost newest event.
fix: should continue to pull data from the leader if EOF on leader read while not disable write
optimize the skipped channel to avoid read disk data
fix: select node by the least load factor. Thanks to @Damon07
fix: handle the possible conflict in the inflight between normal and delayed message
v0.3.7-HA.1.12.4
3 years ago
add API to get a message from the delayed queue and the data tool also support search by the msg-id in the delayed queue.
optimize the delayed queue empty
optimize the topic channel meta save and history pub stats
some optimize for a large number of topics in a single cluster
v0.3.7-HA.1.12.3
3 years ago
add limiter for pub and fail fast if too much waiting
add rate limiter for consuming the large messages
allow setting the max ready for client
wakeup quick if some re-queued messages waiting in the map
optimize and refactor JSON parser
avoid read backend while too much deferred in mem to reduce memory usage
v0.3.7-HA.1.12.2
3 years ago
allow cached the etcd request from the lookup
add some options to control the memory usage
add Prometheus metric for nsqd
timer and ticker leak fix and optimize to improve performance and CPU usage
support disable channel auto-create
optimize the mpub to allow group commit
v0.3.7-HA.1.12.1
3 years ago
optimize the delayed queue sync between ISR
add the restriction on the max delayed queue count
use boltdb for disk write queue meta
add config to always write queue meta to old the file to allow rollback to the old version
use NoFreelistSync to optimize the delayed queue write performance
allow consume committed data on the disabled leader which can speed up the consumer recovery if half of ISR nodes lost.
fix pub timeout timer clean
v0.3.7-HA.1.10.3
3 years ago
add the restriction on the max delayed queue count
2.fix draining channel should check the delayed-type while clear the counter for delayed
3.use Array Type for the boltdb which seems better for delete operation
v0.3.7-HA.1.10.2
3 years ago
fix the delayed message may fail to continue consume if the channel does empty or skip.
v0.3.7-HA.1.10.1
3 years ago
log the trace sub cost time. Thanks to @zhanghuanzhong
force fix reader to queue start if read cursor is invalid while update end
optimize the full sync for delayed queue
optimize consumer latency in some special cases
clean invalid commit log data if init failed with old data
force fix data if start with non-leader or single replica