-
Incident report
-
Resolution: Unresolved
-
Trivial
-
None
-
5.0.8
-
None
Steps to reproduce:
- Setup media type slack if upgrading from previous versions with the official documention: https://git.zabbix.com/projects/ZBX/repos/zabbix/browse/templates/media/slack?at=refs%2Fheads%2Frelease%2F5.0
- Verify on the Zabbix server, that there is no CLOSE_WAIT socket:
root@zabbix:/home/steffen# ss -ano | grep CLOSE-WAIT root@zabbix:/home/steffen#
- Create a new host with an unassigned IP address within your network and add the Template Template Module ICMP Ping
- Wait __ a few minutes, until you receive the alert in Slack
- Check back for CLOSE_WAIT sockets on the Zabbix server:
root@zabbix:/home/steffen# ss -ano | grep CLOSE-WAIT tcp CLOSE-WAIT 40 0 10.10.20.8:47818 3.123.248.34:443 root@zabbix:/home/steffen#
- Create another host (same procedure as before, with another unassigned IP) - or provoke an error otherwise of course.
- Wait again for the alert to show up in Slack
- Check one more time for CLOSE_WAIT sockets
root@zabbix:/home/steffen# ss -ano | grep CLOSE-WAIT tcp CLOSE-WAIT 40 0 10.10.20.8:59914 3.123.248.34:443 tcp CLOSE-WAIT 40 0 10.10.20.8:47818 3.123.248.34:443 root@zabbix:/home/steffen#
- Restart zabbix-server to verify, that CLOSE_WAIT sockets are gone, after the application exits
root@zabbix:/home/steffen# systemctl restart zabbix-server root@zabbix:/home/steffen# ss -ano | grep CLOSE-WAIT root@zabbix:/home/steffen#
Result:
With every new alert a new CLOSE_WAIT socket will spawn and the server will eventually run out of open ports.
Expected:
No leaking sockets/CLOSE_WAIT sockets
Initially I "reported" the issue back in the beginning of 2020 [here|https://www.zabbix.com/forum/zabbix-troubleshooting-and-problems/394351-zabbix-4-4-5-slack-media-type-tcp-close-wait]and here.
I guess nobody is monitoring for CLOSE_WAIT sockets or is not making use of the webhooks.
BTW: The same can be reproduced with Discord as media type.