msg/async/Protocol*: send keep alive if existing wins

This is a follow-up fix of https://github.com/ceph/ceph/pull/25754.

It turns out the existing connection would win the connection race
and then be stuck in the __replacing__ stage forever without being
aware of that the underlying Pipe is actually broken.

Fix by forcing existing sending keepalive periodically __too__.

Fixes: http://tracker.ceph.com/issues/38493
Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
This commit is contained in:
xie xingguo 2019-02-27 15:34:46 +08:00
parent c1112928b7
commit 61b9432ef9
2 changed files with 6 additions and 0 deletions

View File

@ -1998,6 +1998,8 @@ CtPtr ProtocolV1::handle_connect_message_2() {
<< " existing_state="
<< connection->get_state_name(existing->state) << dendl;
reply.global_seq = exproto->peer_global_seq;
// make sure we notice if existing connection is no longer functioning
existing->send_keepalive();
existing->lock.unlock();
return send_connect_message_reply(CEPH_MSGR_TAG_RETRY_GLOBAL, reply,
authorizer_reply);

View File

@ -2464,6 +2464,8 @@ CtPtr ProtocolV2::handle_reconnect(ceph::bufferlist &payload)
ldout(cct, 1) << __func__
<< " existing racing replace happened while replacing."
<< " existing=" << existing << dendl;
// make sure we notice if existing connection is no longer functioning
existing->send_keepalive();
auto retry = RetryGlobalFrame::Encode(session_stream_handlers,
exproto->peer_global_seq);
return WRITE(retry, "session retry", read_frame);
@ -2571,6 +2573,8 @@ CtPtr ProtocolV2::handle_existing_connection(AsyncConnectionRef existing) {
ldout(cct, 1) << __func__
<< " existing racing replace happened while replacing."
<< " existing=" << existing << dendl;
// make sure we notice if existing connection is no longer functioning
existing->send_keepalive();
auto wait = WaitFrame::Encode(session_stream_handlers);
return WRITE(wait, "wait", read_frame);
}