I2 SCS does not recognize GPO endpoint

[lnevers@bbn.com]

ON the I2 SCS all requests that include GPO as an end-point fail with the error:

StitchingServiceFailedError: Error from Stitching Service: code 3:
MxTCE ComputeWorker return error message '
Action_ProcessRequestTopology_MP2P:
:Finish() Cannot find the set of paths for the RequestTopology. '.

The same topologies work if I use the MAX SCS.

Can the installation of the dynamic refresh of Ads be a source of this problem?

[xyang@maxgigapop.net]

There have been bunch of changes in ION and AL2S topologies in past few days. The MAX SCS has relatively older topologies. Getting log from Chad to figure out where the link broke.

[lnevers@bbn.com]

On 12/11/14 3:53 PM, Xi Yang wrote:

I just asked Chad to restart the SCS service.

One catch is that we should always restart SCS after topology update.
SCS will detect topology change in files and try load it up on the fly. 
However, that feature has not worked reliably.

On 12/11/14 5:56 PM, Heidi Picher Dempsey wrote:

Xi--It sounds like you are recommending that the SCS be restarted 
every time there is a topology change, which is not really workable 
for a production SCS, because there will be too many people using it.  
Do you pan to fix the function that detects a topology change and 
loads it on the fly before Jan 6?  If not by then, when?

Until we have a fix, it sounds like Chad needs to write into the 
support procedures for SCS that this restart is needed whenever a 
topology change is requested via the ticket system that was proposed 
for ops in our earlier emails.  We'll probably need to schedule a 
regular weekly maintenance ticket the way we do for portal and 
clearinghouse to make sure there is a time when we CAN do the restarts.

Note: The restart cleared the GPO Endpoint unknown path problem.

[lnevers@bbn.com]

On 12/12/14 8:55 AM, Xi Yang wrote:

It has been part of standard procedure to restart SCS after topology changes in the 
past. Loading updated topology on-the-fly was an experimental feature that I had 
difficulty to make it work stably. 

I will evaluate the amount of work needed to fix it that but I cannot give you a time frame.

SCS is stateless. Restart is very quick. Clients are normally resilient to short 
connection disruptions.  So doing some restarts for now is that that bad. I will 
look into this after winter break.

[lnevers@bbn.com]

This problem was resolved. Closing ticket.