Re: [Qemu-devel] [PATCH v3 10/11] block/nbd-client: nbd reconnect

[Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>]

09.06.2018 18:32, Vladimir Sementsov-Ogievskiy wrote:
> Implement reconnect. To achieve this:
>
> 1. Move from quit bool variable to state. 4 states are introduced:
>     connecting-wait: means, that reconnecting is in progress, and there
>       were small number of reconnect attempts, so all requests are
>       waiting for the connection.
>     connecting-nowait: reconnecting is in progress, there were a lot of
>       attempts of reconnect, all requests will return errors.
>     connected: normal state
>     quit: exiting after fatal error or on close
>
> Possible transitions are:
>
>     * -> quit
>     connecting-* -> connected
>     connecting-wait -> connecting-nowait
>     connected -> connecting-wait
>
> 2. Implement reconnect in connection_co. So, in connecting-* mode,
>      connection_co, tries to reconnect every NBD_RECONNECT_NS.
>      Configuring of this parameter (as well as NBD_RECONNECT_ATTEMPTS,
>      which specifies bound of transition from connecting-wait to
>      connecting-nowait) may be done as a follow-up patch.
>
> 3. Retry nbd queries on channel error, if we are in connecting-wait
>      state.
>
> 4. In init, wait until for connection until transition to
>      connecting-nowait. So, NBD_RECONNECT_ATTEMPTS is a bound of fail
>      for initial connection too.
>
> Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>

squash:

@@ -616,7 +617,10 @@ static coroutine_fn int nbd_co_receive_one_chunk(
          s->reply.handle = 0;
      }

-    if (s->connection_co) {
+    if (s->connection_co && !s->wait_in_flight) {
+        /* We must check s->wait_in_flight, because we may entered by
+         * nbd_recv_coroutines_wake_all(), int this case we should not
+         * wake connection_co here, it will woken by last request. */
          aio_co_wake(s->connection_co);
      }



-- 
Best regards,
Vladimir