Let user configure default timeouts on a per-RPC basis #29018
Labels
module: rpc
Related to RPC, distributed autograd, RRef, and distributed optimizer
triaged
This issue has been looked at a team member, and triaged and prioritized into an appropriate module
馃殌 Feature
#28392 is adding timeouts per RPC, but it is currently a global timeout. Per @pritamdamania87's suggestion, we should:
1.Have a timeout parameter in RpcAgent.send().
2. For user RPCs, we always fill that in with the default rpc timeout.
3. For system RPCs, it'll default to 0 (which would be infinite) unless the system RPC sets it.
4. Later we can allow the user level RPC API to specify timeouts as well.## Motivation
cc @pietern @mrshenli @pritamdamania87 @zhaojuanmao @satgera @gqchen @aazzolini @rohan-varma @xush6528
The text was updated successfully, but these errors were encountered: