Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Trainer send term signal #11220

Merged
merged 13 commits into from
Jun 12, 2018

Conversation

typhoonzero
Copy link
Contributor

Fix #11077

Call exe.executor.complete() to tell pserver to mark current trainer as finished.

gongweibao
gongweibao previously approved these changes Jun 11, 2018
Copy link
Contributor

@gongweibao gongweibao left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@@ -52,6 +52,9 @@ bool RequestSendHandler::Handle(const std::string& varname,
if (varname == BATCH_BARRIER_MESSAGE) {
VLOG(3) << "sync: recv batch barrier message";
rpc_server_->IncreaseBatchBarrier(kRequestSend);
} else if (varname == COMPLETE_MESSAGE) {
VLOG(3) << "sync: recv complete message";
rpc_server_->DecreaseClientNum();
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Smart!

@gongweibao
Copy link
Contributor

Only number maybe not stable. Do we need to change them to tolerant with client's multiple COMPLETE_MESSAGE BATCH_BARRIER_MESSAGE and so on in one barrier?

@typhoonzero
Copy link
Contributor Author

@gongweibao You were right, we may need to add trainer id to the request when we want to make all RPC calls retriable.

Copy link
Contributor

@gongweibao gongweibao left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@typhoonzero typhoonzero merged commit 34865f2 into PaddlePaddle:develop Jun 12, 2018
@typhoonzero typhoonzero deleted the trainer_send_term_signal branch June 12, 2018 08:09
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants