Re: [libqb] libqb - Did you come right?
On Tue, Dec 14, 2010 at 07:54:08AM +0000, Dietmar Maurer wrote:
Hi Angus,
> Had you come right with the problems you were having?
Please apply the following patch to the example test server:
--- bms.c.org 2010-12-14 08:40:23.000000000 +0100
+++ bms.c 2010-12-14 08:41:11.000000000 +0100
@@ -127,7 +127,7 @@
perror("qb_ipcs_event_send");
}
}
- return 0;
+ return -1;
}
After recompile, the server continues to work - 'bmc' runs quite normal.
I thought it should be removed from the event loop if s1_msg_process_fn returns
something < 0?
We that's not how it is currently designed.
Lest say you have hundreds of async requests comming in. Now you
have a big input queue, to speed things up the IPC server processes
a bunch of requests before going back to the poll loop. If during
this you start having trouble processing them all - then you
return -1 from your process_message function. This will cause
the IPC server to stop doing bulk processing and go back to the
poll loop.
If you want to block normal incomming requests use:
qb_ipcs_request_rate_limit(s, QB_IPCS_RATE_OFF);
then later to turn it on again
qb_ipcs_request_rate_limit(s, QB_IPCS_RATE_NORMAL);
Or to shutdown the connection.
qb_ipcs_disconnect(c);
The second problem is a bit artificial - don't know if we really need to handle that.
When
you apply the following patch:
--- bms.c.org 2010-12-14 08:40:23.000000000 +0100
+++ bms.c 2010-12-14 08:49:13.000000000 +0100
@@ -113,6 +113,9 @@
response.size = sizeof(struct qb_ipc_response_header);
response.id = 13;
response.error = 0;
+
+ return 0;
+
if (blocking) {
res = qb_ipcs_response_send(c, &response,
sizeof(response));
The client (bmc) hangs forever, Well, the server misbehaves and never send a response. I
just
thought a timeout on the client side would be a good idea?
Are you trying the latest? try "git pull".
I did the following:
diff --git a/tests/bmc.c b/tests/bmc.c
index 5eafd56..065cf5b 100644
--- a/tests/bmc.c
+++ b/tests/bmc.c
@@ -103,6 +103,7 @@ repeat_send:
res = qb_ipcc_recv(conn,
&res_header,
sizeof(struct qb_ipc_response_header));
+ printf("qb_ipcc_recv: %d\n", res);
if (res == -EAGAIN) {
goto repeat_recv;
}
diff --git a/tests/bms.c b/tests/bms.c
index c8a6c3b..85ad02e 100644
--- a/tests/bms.c
+++ b/tests/bms.c
@@ -105,6 +105,8 @@ static int32_t s1_msg_process_fn(qb_ipcs_connection_t *c,
struct qb_ipc_response_header response;
ssize_t res;
+ return 0;
+
if (verbose > 2) {
printf("%s:%d %s > msg:%d, size:%d\n",
__FILE__, __LINE__, __func__,
when I run this i get tons of:
qb_ipcc_recv: -11
qb_ipcc_recv: -11
qb_ipcc_recv: -11
At which you can handle backing off or disconnecting.
Regards
-Angus
- Dietmar