Details
-
Sub-task
-
Status: Closed
-
Minor
-
Resolution: Duplicate
-
0.9.2-incubating
-
None
Description
In my test topology, each worker has 108 threads. Some threads are generated from a thread pool with a vague name.
For example, the netty threads, the worker receiver thread, the worker transfer threads, the disruptor thread are not good named.
May be we should name them with a more meaningful name so that it is more easy for us to track the performance issues.
Here is the naming I used:
name description
-------------------------------------
worker-transfer-thread: the single transfer thread for each worker
worker-receiver-thread: the receiver thread for each worker
netty-server-boss-host-port-seqenceId: the netty server boss thread.
netty-server-worker-host-port-sequenceId:the netty server worker thread
executor-disruptor-transfer-thread-[executorId]: executor transfer thread for disrupotor queue.
netty-client-boss-targetHost-port-sequenceId: the netty client bos thread
netty-client-worker-targetHost-port-sequenceId: the netty client worker thread.
Attachments
Issue Links
- is duplicated by
-
STORM-297 Storm Performance cannot be scaled up by adding more CPU cores
- Resolved