Carsten_Schmitz

Race Condition with two subsequent JOBF

Discussion created by Carsten_Schmitz on Jul 27, 2017
Latest reply on Aug 2, 2017 by Carsten_Schmitz
Hi.

We noticed there appears to be a race condition with subsequent JOBF, especially if high latency network connections are involved. I can quite reliably (about 50% of the time) replicate the problem. The setup is simply two subsequent JOBF transfering a test file of about a megabyte in size, like this:

Host #1 ---> JOBF over high-latency network link ---> Host #2 ---> JOBF ---> Host #3

What happens then is this: The second JOBF starts eventhough the file has not yet been fully transfered to HOST #2 by the first JOBF, causing the second JOBF (and the entire JOBP) to hang indefinetly.

(edit): If one inserts a sleep statement of five seconds in between the two JOBF, the problem entirely disappears, proving to me this is indeed a timing issue and at the core one of lacking serialization in the execution of objects.

I opened a support case for this, but since we're on version 10 and this is not deemed critical enough to warrant handling as a critical maintainance issue, a problem report will not be created at this time.

Thus, I will revisit this once we upgrade to version 12, but out of curiosity: Has anyone else seen this behaviour, and if so, with which AE version?

Cheers,
Carsten

Outcomes