Reputation: 139
We have J2EE application (deployed in wildfly) which is deployed in Multiple Data Centre in different region (UK and Australia). Application deployed in UK is making HTTPS REST API call (using HttpClient) to application deployed in Australia.
Normally it works fine. But when connection is slow between UK and Australia, CPU of UK application reaches to 100 percentage.
We have taken thread dump and it shows that many threads (which are making HTTPS REST CALL) are RUNNABLE status. What could reason for that? How can fix the problem?
Upvotes: 0
Views: 1542
Reputation: 53
Explain it by example:
Let say the UK application receive 1000 calls per secs.
When the connection between UK and Australia is normal. It might take 1 sec to process 1000 calls.
When the connection between UK and Australia is slow. It might take 3 secs to process 1000 calls. In between this 3 secs, (the UK application receive 1000 calls per secs), so i will have total 3000 calls pending at UK application.
This make the CPU become 100%
Upvotes: 1