latchset
Tang doesn't close socket if error occured?
An unpleasant surprise was the inability to manage services via systemd and the high workload of the server. When trying to do something with the services, I got the error: Failed to load daemon: No buffer space available.
It turned out that there are 77 334 systemd units in the system of the form tangd@53241-<local ip>:80-<remote ip>:49204.service.
# systemctl list-units --all | grep tang
...
● tangd@9993-<local ip>:80-<remote ip>:41608.service loaded failed failed Tang Server
● tangd@9994-<local ip>:80-<remote ip>:41642.service loaded failed failed Tang Server
● tangd@9995-<local ip>:80-<remote ip>:41648.service loaded failed failed Tang Server
● tangd@9996-<local ip>:80-<remote ip>:41654.service loaded failed failed Tang Server
● tangd@9997-<local ip>:80-<remote ip>:41664.service loaded failed failed Tang Server
● tangd@9998-<local ip>:80-<remote ip>:41670.service loaded failed failed Tang Server
...
If I check log via journalctl -xeu tangd@63-...:62934.service I see this:
May 04 16:44:16 test-node-2 tangd[29047]: HTTP Parsing Error: invalid HTTP method
Hello @MAH69IK
Can you please provide next information?:
- tang version
- operating system
- steps to reproduce the issue
I have tested in Fedora and RHEL 9.1, with latest versions of tang in both of them, and I don't see this issue. I am starting/stopping tang by using:
#systemctl enable --now tangd.socket #systemctl disable --now tangd.socket
Hello!
OS: Debian 10 Tang: 7-1+deb10u1 (latest for this OS)
Steps to reproduce: well, I just started tang (tangd.socket) and wait few days.
Hello @MAH69IK . Could you please try downgroading libhttp-parser to see if that helps, as described in #79?