cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
1230
Views
0
Helpful
1
Replies

CWA Webservice Timeouts

Cale Montgomery
Level 1
Level 1

We're hitting an odd issue with json app file transfer.  It works... most of the time... unless the transfer takes longer than 5 minutes.  This causes connection to be reset and the job to abend.  Files have been transferred, but we would like for the connection to be maintained for the entirety of the job.

 

Looking at the job definition using the CWA web browser, we have the Timeout value under the Web Service Options tab set at 1800 seconds.  Looking at the same tab using the Java client, we're locked into 75 seconds.  Making a change to this value does not actually persist on saving the job.  It always shows 75 seconds.

 

Looking at the same tab in the Web Service Connection definition, we have 1800 seconds for both the Web Client and Java.

 

But, still, our connection always resets at 5 minutes.  That matches neither the 1800 seconds, or the 75 seconds.  Can someone offer advice on how to allow this connection to persist beyond 5 minutes?

 

Thank you.

1 Accepted Solution

Accepted Solutions

Cale Montgomery
Level 1
Level 1

Answered my own question.  5 minutes is our default TCP/IP idle timeout.  The REST call from CWA is doing nothing but initiating a SOAP call from a second internal server, and then going idle.

 

While I've not been able to get a satisfactory answer on why we don't just launch the SOAP call from CWA itself, I have managed to convince the devs who engineered this process to exchange a keep-alive down the REST connection until the process has completed.

View solution in original post

1 Reply 1

Cale Montgomery
Level 1
Level 1

Answered my own question.  5 minutes is our default TCP/IP idle timeout.  The REST call from CWA is doing nothing but initiating a SOAP call from a second internal server, and then going idle.

 

While I've not been able to get a satisfactory answer on why we don't just launch the SOAP call from CWA itself, I have managed to convince the devs who engineered this process to exchange a keep-alive down the REST connection until the process has completed.