|
If you were logged in you would be able to see more operations.
|
|
|
|
File Attachments:
|
|
|
Issue Links:
|
Incorporates
|
|
|
|
This issue is part of:
|
|
|
HADOOP-2864 Improve the Scalability and Robustness of IPC
|
|
|
|
|
|
|
| Hadoop Flags: |
Reviewed, Incompatible change
|
| Release Note: |
Replaced timeouts with pings to check that client connection is alive. Removed the property ipc.client.timeout from the default Hadoop configuration. Removed the metric RpcOpsDiscardedOPsNum.
Replaced timeouts with pings to check that client connection is alive. Removed the property ipc.client.timeout from the default Hadoop configuration. Removed the metric RpcOpsDiscardedOPsNum.
|
| Resolution Date: |
05/May/08 09:27 PM
|
|
Current RPC (really IPC) relies on client side timeouts to find "dead" sockets. I propose that we have a thread that once a minute (if the connection has been idle) writes a "ping" message to the socket. The client can detect a dead socket by the resulting error on the write, so no client side timeout is required. Also note that the ipc server does not need to respond to the ping, just discard it.
|
|
Description
|
Current RPC (really IPC) relies on client side timeouts to find "dead" sockets. I propose that we have a thread that once a minute (if the connection has been idle) writes a "ping" message to the socket. The client can detect a dead socket by the resulting error on the write, so no client side timeout is required. Also note that the ipc server does not need to respond to the ping, just discard it. |
Show » |
| No work has yet been logged on this issue.
|
|
Improvement
Closed
RPC should send a ping rather than use client timeouts
