[
https://issues.redhat.com/browse/TEIID-5962?page=com.atlassian.jira.plugi...
]
Steven Hawkins commented on TEIID-5962:
---------------------------------------
And I think that behavior need to improve. You can try to look at
Netflix Hystrix as example of this logic.
There are several take-aways here.
1. Teiid Spring Boot not using a connection pool with a resource that can consume a thread
could result in too much resource consumption in a scenario like this. There is already
captured as TEIIDSB-163
2. It is not clear from what you have shown above what threads exactly are being held. A
thread dump would help clarify. As Ramesh is saying if it's something under the
salesforce library it will take some investigation as to how/why to interrupt/cleanup that
thread. The Teiid engine already has quite a bit of logic to make sure that it's
threads are managed appropriately.
3. The main doc set should show the timeout related properties.
I'll at least work #3 under this issue. For #2 if you can provide a thread dump that
would be great.
Teiid locks many threads when Salesforce is disconnected
--------------------------------------------------------
Key: TEIID-5962
URL:
https://issues.redhat.com/browse/TEIID-5962
Project: Teiid
Issue Type: Bug
Components: Salesforce Connector
Affects Versions: 12.3.1
Reporter: Renat Eskenin
Assignee: Steven Hawkins
Priority: Major
Attachments: Снимок экрана от 2020-05-31 11-09-10.png
We had two incidents in our experimental service with teiid spring boot app. When
Salesforce is down to 20min and service with teiid trying to call SQL from Salesforce we
have many threads in tomcat server (in spring boot app). Then when 200 threads is
activated service do not reply to requests because is "max threads" in tomcat of
service.
Why teiid do not send response as "Salesforce is down - error" or connection
timeout?
Micro schema:
HTTP Requests to REST api -> Teiid spring boot app -> {broken 20min}Salesforce
--
This message was sent by Atlassian Jira
(v7.13.8#713008)