Tranquil IT Forum

Hello,

Since the middle of last week, nearly a third of our machines have been showing as disconnected in WAPT, although they are accessible from the network.

In the agent logs, I find the following:

2019-04-01 13:49:16,072 WARNING Websocket connect params: HTTPSConnectionPool(host='srv00-wapt01', port=443): Max retries exceeded with url: /get_websocket_auth_token (Caused by NewConnectionError('
2019-04-01 13:50:16,082 WARNING Websocket connect params: HTTPSConnectionPool(host='srv00-wapt01', port=443): Max retries exceeded with url: /get_websocket_auth_token (Caused by NewConnectionError('
2019-04-01 13:51:16,095 WARNING Websocket connect params: HTTPSConnectionPool(host='srv00-wapt01', port=443): Max retries exceeded with url: /get_websocket_auth_token (Caused by NewConnectionError('
2019-04-01 13:52:16,108 WARNING Websocket connect params: HTTPSConnectionPool(host='srv00-wapt01', port=443): Max retries exceeded with url: /get_websocket_auth_token (Caused by NewConnectionError('
2019-04-01 13:53:16,121 WARNING Websocket connect params: HTTPSConnectionPool(host='srv00-wapt01', port=443): Max retries exceeded with url: /get_websocket_auth_token (Caused by NewConnectionError('
2019-04-01 13:54:16,134 WARNING Websocket connect params: HTTPSConnectionPool(host='srv00-wapt01', port=443): Max retries exceeded with url: /get_websocket_auth_token (Caused by NewConnectionError('
2019-04-01 13:55:16,970 WARNING Websocket connect params: HTTPSConnectionPool(host='srv00-wapt01', port=443): Max retries exceeded with url: /get_websocket_auth_token (Caused by NewConnectionError('

How can I fix this?

Version 1.7.3.5 installed on a Windows Server 2016 (1607).

Thank you for your help.

Hello,

I've tried to make some progress on this issue.
It seems the agents are opening many sockets to the server, and at a certain point, there are no more ports available for communication. I thought this was due to the listening ports on the server, but it seems to be on the client machines.
I installed a new WAPT server on CentOS 7 and I have the same problem with 3 out of 5 machines communicating with this new server...

According to the error message (10055) and numerous forum posts, it seems that Python (or Windows) isn't closing the connections correctly and that the OS quickly exhausts the available ports.

Have you ever encountered this problem?

Best regards,

Hello,
I don't recall seeing this problem before...
It would be interesting to see why the `/get_websocket_auth_token` request to the server is initially refused (and potentially improperly released, leaving an open socket).

In a command prompt with elevated privileges, run `runwaptservce.bat -ldebug` and observe the behavior.
Check for any interception of HTTPS connections by a local or remote antivirus program.

You can also run `netstat -a -n` on the client to see all the sockets and verify that there are far too many connected to the WAPT server on port 443.

Normally, there is one socket permanently in ESTABLISHED mode (the client's WebSocket connection to the server for actions from the console) and others possibly in TIME_WAIT mode from updates or other previous requests.
Sockets in TIME_WAIT or CLOSE_WAIT mode should not remain open for more than four minutes.

Hello,

I'm having trouble providing you with service logs.
When the problem occurs and we run the command, it restarts the service and it works... I'll try leaving the debugger running for several days on a machine.

Regards,