Project

General

Profile

Bug #9182

stop_disposition and a missing procedure results in 'No export or no access to'

Added by Galya B almost 2 years ago. Updated almost 2 years ago.

Status:
Test
Priority:
Normal
Assignee:
-
Target version:
-
Start date:
Due date:
% Done:

100%

billable:
No
vendor_id:
GCD
case_num:
version_reported:
version_resolved:
reviewer:
production:
No
env_name:
topics:

Related issues

Related to Base Language - Feature #4065: server-side processing of client platform dependencies Closed

History

#2 Updated by Galya B almost 2 years ago

Reported in #8964-88 with steps to recreate:

  • use stop_disposition=1 in directory.xml
  • run a FWD client with a missing.p (the program really does not exist) - this will re-login the same FWD client and get you the No export or no access to exception

Constantin, what does 're-login' mean? Is SSO enabled?

#3 Updated by Greg Shah almost 2 years ago

  • Project changed from Bugs to Runtime Infrastructure

#4 Updated by Constantin Asofiei almost 2 years ago

SSO is not involved. Just start a server with stop_disposition=1 and try to runa ChUI client with a missimg program.

The client will try to re-initiliaze and this is when it fails.

#5 Updated by Galya B almost 2 years ago

  • Related to Feature #4065: server-side processing of client platform dependencies added

#6 Updated by Galya B almost 2 years ago

So, here's the story: There was a mechanism in OSResourceManager (initialize) used by ClientSideResourceManager to "initialize a resource only once per JVM", and I've actually believed this and reworked the code to be initialized only once per JVM, missing the fact that ThinClient.initOSResources explicitly initializes certain rpc interfaces multiple times with stop_disposition of 1.

The good news is that noone else will be misled in the future. The bad news is that there is a bug, but we already know this. The fix will be merged with 4065g.

#7 Updated by Galya B almost 2 years ago

Constantin, there probably needs to be a max attempt number for the recreate, because with a missing procedure the loop never stops.

#8 Updated by Galya B almost 2 years ago

  • Status changed from New to WIP

#9 Updated by Galya B almost 2 years ago

  • Status changed from WIP to Review
  • % Done changed from 0 to 100

Fix committed to 4065g r15453.

#10 Updated by Greg Shah almost 2 years ago

I agree we should not get into an infinite loop in this case (missing startup procedure).

We should detect it and end the session the first time it is raised. Checking multiple times will never find a different outcome.

#11 Updated by Galya B almost 2 years ago

Greg Shah wrote:

I agree we should not get into an infinite loop in this case (missing startup procedure).

We should detect it and end the session the first time it is raised. Checking multiple times will never find a different outcome.

Fix in 4065g r15455.

#12 Updated by Galya B almost 2 years ago

  • Status changed from Review to Test

4065g was merged to trunk as rev. 15484 and archived.

Also available in: Atom PDF