[PLC4X-139] PLC4X leaks sockets in case of connection problems

• Type: Bug
• Status: Closed
• Priority: Blocker
• Resolution: Fixed
• Affects Version/s: 0.4.0
• Fix Version/s: 0.5.0
• Component/s: Core
• Assignee: Julian Feinauer
• Reporter: Julian Feinauer
• Created: 02/Aug/19 15:12
• Updated: 03/Dec/19 15:35
• Resolved: 23/Sep/19 15:19

Description

See my description on the list: https://lists.apache.org/thread.html/4aa229421668a92275e33643dbc57f2fea29cee7ffcb07891f58c9eb@%3Cdev.plc4x.apache.org%3E

I created a MWE like that

public static void main(String[] args) throws Exception {
        IntStream.range(0, 100).parallel().forEach(i -> createConnection(i));
    }

    private static void createConnection(int i) {
        System.out.println("trying " + i);
        try (PlcConnection plcConnection = new PlcDriverManager().getConnection("s7://192.168.167.211/0/0")) {

        } catch (Exception e) {
         logger.error("Kann keine Verbindung zu der angegebenen Adresse aufbauen");
        }
    }

where the IP is not reachable, thus it fails every time.

When looking at the open sockets with (works on my Mac)

while true; do lsof -p 16843 | wc -l; sleep 10; done

I get the output

    1640
    1640
    1640
    1736
    1736
    1736
    1832
    1832
    1832
    1928
    1928
    1928
    2024
    2024
    2024
    2120
    2120
    2120
    2216
    2216
    2216
    2312
    2312
    2312
    2408
    2408
    2408
    2504
    2504
    2504
    2500

So this reall sems to be the case.

I think this could be due to us only propagating the exception but not closing the channel, I will investigate further.

Issue Links

• relates to: PLC4X-158 Too many open files after less than 24 hours of Scraper use
• links to: GitHub Pull Request #76