Hello,
I made a mistake in my last message. Please see my inline comment for
the correction
Cheers,
Cliff
-----Original Message-----
From: Clifford THOMPSON [mailto:[EMAIL PROTECTED]
Sent: February 22, 2008 15:08
To: Apache AXIS C Developers List
Subject: RE: FW: Question regarding the adjustment of response timeouts
Hello Samisa,
I have been doing some tracing, and hopefully I've found some
information that may be useful. I have traced through my blocking
service call:
+ axis2_stub_op_MyService_MyOperation
+ axis2_svc_client_send_receive_with_op_name
+ axis2_op_client_execute
+ axis2_op_client_two_way_send
It appears that the call to "axis2_msg_ctx_get_property" returns NULL:
<code>
AXIS2_EXTERN axis2_msg_ctx_t *AXIS2_CALL
axis2_op_client_two_way_send(
const axutil_env_t * env,
axis2_msg_ctx_t * msg_ctx)
{
axis2_engine_t *engine = NULL;
axis2_status_t status = AXIS2_SUCCESS;
axis2_msg_ctx_t *response = NULL;
axis2_conf_ctx_t *conf_ctx = NULL;
axis2_op_t *op = NULL;
axiom_soap_envelope_t *response_envelope = NULL;
axutil_property_t *property = NULL;
long index = -1;
axis2_bool_t wait_indefinitely = AXIS2_FALSE;
AXIS2_ENV_CHECK(env, NULL);
conf_ctx = axis2_msg_ctx_get_conf_ctx(msg_ctx, env);
engine = axis2_engine_create(env, conf_ctx);
if (!engine)
return NULL;
property =
axis2_msg_ctx_get_property(msg_ctx, env,
AXIS2_TIMEOUT_IN_SECONDS); </code>
And thus the next section of code never picks up the timeout value
since we never find the "property" for the timeout:
<code>
if (property)
{
axis2_char_t *value = axutil_property_get_value(property,
env);
if (value)
index = AXIS2_ATOI(value);
if (index == -1)
{
wait_indefinitely = AXIS2_TRUE;
index = 1;
}
}
</code>
which is used later in the function:
<code>
else
{
while (!response_envelope && index > 0)
{
/*wait till the response arrives */
AXIS2_SLEEP(1);
if (!wait_indefinitely)
index--;
response_envelope =
axis2_msg_ctx_get_response_soap_envelope(msg_ctx,
env);
}
</code>
You guys are much more familiar which the code base, so does the call
to "axis2_op_client_get_msg_ctx" copy the timeout information from the >
"options" member of "op_client" into "msg_ctx" (the variable that is
eventually passed to "axis2_msg_ctx_get_property"):
I meant the call to "axis2_msg_ctx_set_options" not
"axis2_op_client_get_msg_ctx", please see the additional code in the
snippet below:
<code>
AXIS2_EXTERN axis2_status_t AXIS2_CALL
axis2_op_client_execute(
axis2_op_client_t * op_client,
const axutil_env_t * env,
const axis2_bool_t block)
{
axis2_conf_ctx_t *conf_ctx = NULL;
axis2_msg_ctx_t *msg_ctx = NULL;
axis2_transport_out_desc_t *transport_out = NULL;
axis2_transport_in_desc_t *transport_in = NULL;
axis2_status_t status = AXIS2_FAILURE;
axis2_op_t *op = NULL;
axis2_char_t *msg_id = NULL;
AXIS2_ENV_CHECK(env, AXIS2_FAILURE);
if (op_client->completed)
{
return AXIS2_FAILURE;
}
conf_ctx = axis2_svc_ctx_get_conf_ctx(op_client->svc_ctx,
env);
msg_ctx = (axis2_msg_ctx_t *)
axis2_op_client_get_msg_ctx(op_client, env,
AXIS2_WSDL_MESSAGE_LABEL_OUT);
</code>
<code>
if (!msg_ctx)
{
return AXIS2_FAILURE;
}
axis2_msg_ctx_set_options(msg_ctx, env, op_client->options);
</code>
When I traced through the call to "axis2_msg_ctx_get_property", all of
the hash tables stored within "msg_ctx" or its children appear to be > >
empty. I am not sure if this is correct or not, but I hope this all
helps out with the bug tracking.
Cheers,
Cliff
-----Original Message-----
From: Samisa Abeysinghe [mailto:[EMAIL PROTECTED]
Sent: February 20, 2008 18:06
To: Apache AXIS C Developers List
Subject: Re: FW: Question regarding the adjustment of response
timeouts
The current problem is that, for the normal non blocking case, the
implementation does not pick the user option and set it on the socket.
I hope the fix is straight forward.
Regards,
Samisa...
Clifford THOMPSON wrote:
I meant 60000ms...
-----Original Message-----
From: Clifford THOMPSON [mailto:[EMAIL PROTECTED]
Sent: February 20, 2008 13:40
To: Apache AXIS C Developers List
Subject: RE: FW: Question regarding the adjustment of response
timeouts
Hello Senaka,
I took a look at "axis2_http_transport.h" and noticed that the
constants, AXIS2_HTTP_DEFAULT_SO_TIMEOUT and
AXIS2_HTTP_DEFAULT_CONNECTION_TIMEOUT, both held values of 6000ms.
This coincides with the upper timeout limit our team was
experiencing,
so it may provide a clue to the timeout problem.
Cheers,
Cliff
-----Original Message-----
From: Senaka Fernando [mailto:[EMAIL PROTECTED]
Sent: February 18, 2008 11:35
To: Apache AXIS C Developers List
Subject: Re: FW: Question regarding the adjustment of response
timeouts
Hi Cliff,
We'll look into this before the 1.3.0 release and try to have it
fixed
before we release.
Regards,
Senaka
Hello Dev Team,
I presented this question with regards to using timeouts in the
axis2-c-user forum. Dimuthu is getting similar results under Linux,
and suggested that there may be a bug in timeout behaviour. Please
see
below for the details.
Cheers,
Cliff
-----Original Message-----
From: Dimuthu Gamage [mailto:[EMAIL PROTECTED]
Sent: February 16, 2008 05:23
To: Apache AXIS C User List
Subject: Re: [AXIS2C] Question regarding the adjustment of response
timeouts
Hi,
I too checked it in Linux and got the same result,
Seems we are not using axis2_options_get_timeout_in_milli_seconds
anywhere.. If this is a bug, should be fixed before the 1.3
release.
Thanks
Dimuthu
On Feb 16, 2008 1:07 AM, Clifford THOMPSON
<[EMAIL PROTECTED]> wrote:
Hello,
I have a question about adjusting the timeout period for web
services.
Our current software dictates that we can have upwards of a 300
second
delay before a response is sent (we have a large amount of data
that
needs to be prepared before being sent). Currently, our web
service
component will timeout after roughly 60 sec (I'm not sure if this
is
the Axis API, or from the OS). I have tried using some of the
timeout
functions in the Axis2C API, but they appear to have no effect (if
I
set the timeout 5 secs and the server takes 10 secs to respond,
the
client will wait 10 secs for the response). I am assuming that I
am
using the API incorrectly. We are working under WinXP, and have
generate portions of our code using the WSDL2C tool.
We have chosen to generate synchronous code using WSDL2C (so the
eventual call in the generate code will be to
"axis2_svc_client_send_receive_with_op_qname"). Here is a rough
paraphrase of the code that we have and how I thought the timeout
function should be
applied:
env = axutil_env_create_all( "MyServiceLog.log",
AXIS2_LOG_LEVEL_TRACE);
assert(NULL != env);
stub = axis2_stub_create_MyService( env,
AXIS2_GETENV("AXIS2C_HOME"),
"http://myserver.ca:8080/services/MyService";);
assert(NULL != stub);
status = axis2_options_set_timeout_in_milli_seconds(
axis2_stub_get_options( stub,
env ),
env,
300000);
assert(AXIS2_SUCCESS == status);
/* */
/* lots of interleaving non-Axis2C code */
/* */
responseNode = axis2_stub_op_MyService_MyOperation(
stub,
env,
headerNode1,
headerNode2,
bodyNode);
if(NULL !=)
{
/* process the response */
}
else
{
/* log the response error */
}
Thank you in advance for the help.
Cheers,
Cliff
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
