InfluxDB 1.8.9 no longer start correctly with systemd on EL 8


after updating to influxdb 1.8.9 i am no longer able to start influxdb with systemd correctly, systemctl start influxd leads to an error. I have this issue on multiple hosts.

[root@infdb01 influxdb]# systemctl start influxd
Job for influxdb.service failed because the control process exited with error code.
See “systemctl status influxdb.service” and “journalctl -xe” for details.

with journalctl -xe i can see:[55534]: Failed to reach influxdb https endpoint at https://localhost:8086/health
systemd[1]: influxdb.service: Control process exited, code=exited status=1[55534]: ts=2021-08-19T16:51:13.096993Z lvl=info msg=“Signal received, initializing clean shutdown…” log_id=0W3q0Vk0000[55534]: ts=2021-08-19T16:51:13.097018Z lvl=info msg=“Waiting for clean shutdown…” log_id=0W3q0Vk0000[55534]: [tcp] 2021/08/19 18:51:13 tcp.Mux: Listener at failed failed to accept a connection, closing all listeners >[55534]: ts=2021-08-19T16:51:13.097154Z lvl=info msg=“Shutting down monitor service” log_id=0W3q0Vk0000 service=monitor

After this, influxd is still running but only serve opentsdb on port :4242 and all other ports are closed.

Starting influxdb by hand by switching to user influxdb an type “influxd --config /etc/influxdb/influxdb.conf” works, all ports open and influxdb working as expected.

Any idea what is causing this? its really annoying to start influxdb by hand after every system update.

Hi tineidae,
had this issue on a 1.8.9 version on a raspbian buster system, too.
I changed the process start-up type from “forking” to “exec” in /etc/systemd/system/influxd.service.
Simply commenting out “Type=forking” and thus turning it to a type of “simple” will work, too.
systemctl daemon-reload && systemctl restart influxdb and that fixed it for me.


1 Like

See startup script stuck in while loop if http auth set · Issue #21967 · influxdata/influxdb · GitHub for bug report

1 Like

Hi DeLaCheese,

i changed systemd on three installations and all of them work like a charm now.

Thanks alot!

Looks like these issues still persist :frowning: I am not comfortable with changing startup scripts, since we are deploying many machines and further upgrades will likely revert my changes.