InfluxDB API unavailable after X attempts...

[thoduh]

Suddenly InfluxDB 1.8.10 does not run anymore :-(

Everything I have found here so far has unfortunately not helped.
What can I do?

[thoduh]

After I have changed the influxd-systemd-start.sh to old Version 1.8.7, no "InfluxDB API unavailable after X attempts..." more.
But port 8086 still causes problems.
I don't understand it, shortly after starting Influxdb 8086 is there, then suddenly not anymore.
Can no one help me further?

[tarundixitravi]

Hi same issue,

Did able to solve it ?

[thoduh]

Upgrade to Version 2.4…

[nocturno66]

Hi, I am in the same situation.
Is the upgrade the solution?

Thanks

[thoduh]

Yes, so it worked for me 😉

[nocturno66]

Ok, thanks. I will try

[ahpnils]

I recently came across the same issue, following a simple reboot of an InfluxDB instance I manage. This is a fairly large instance I think, our influx data directory is about 150GB. It turned out our instance needed more than 24 minutes to start. With a workmate, we noticed that even with those "InfluxDB API unavailable after X attempts" lines, we could see database files still being loaded. We also saw that after a loooong time, either the service started, either systemd killed it.

We ended up overriding some systemd settings, especially a timeout value :

$ cat /etc/systemd/system/influxdb.service.d/override.conf
[Service]
TimeoutStartSec=45m

Our file contains more than that, but this is the only setting that matters in this case. And a previous colleague had set this up with 15 minutes. After this override file creation or update, don't forget to run systemctl daemon-reload as root or with sudo for changes to be applied.

Hope this helps !

[AnnaWojciechowska]

In some cases it can be avialble RAM. I had a machine with 2 GB and when I added data suddenly influx needed 2.5 and it would just fail to restart, got timeout errors, but actually adding memory helped - works now, but consumes 2.5 GB - I run old 1.8 influxd, hopefully newer versions are less memory greedy

[masterbzchief]

I am running influxdb and storing mqtt data from a node-red flow. I've tried commenting out the Type=forked, ran daemon-reload and it did not fix this issue. What did work, was going in /etc/systemd/system and ensuring that the node-red service start after influxdb (After=influxdb.service).

[averter]

I recently came across the same issue, following a simple reboot of an InfluxDB instance I manage. This is a fairly large instance I think, our influx data directory is about 150GB. It turned out our instance needed more than 24 minutes to start. With a workmate, we noticed that even with those "InfluxDB API unavailable after X attempts" lines, we could see database files still being loaded. We also saw that after a loooong time, either the service started, either systemd killed it.

We ended up overriding some systemd settings, especially a timeout value :

$ cat /etc/systemd/system/influxdb.service.d/override.conf
[Service]
TimeoutStartSec=45m

Our file contains more than that, but this is the only setting that matters in this case. And a previous colleague had set this up with 15 minutes. After this override file creation or update, don't forget to run systemctl daemon-reload as root or with sudo for changes to be applied.

Hope this helps !

Thank you very much for this helpful tip. For my RPI3 I had to add that line in a different file using sudo nano /etc/systemd/system/influxd.service and then sudo systemctl status influxdb after the sudo systemctl daemon-reload. Worked like a charm.

[melroy89]

For anybody, including my self. I enabled socket file in InfluxDB and this require that the service has access to /var/run/influxdb.

Every time I update influxdb, I need to add the following 2 lines again to the influxdb.service under Ubuntu server:

[Service]
....

RuntimeDirectory=influxdb
RuntimeDirectoryMode=0755

[SEGELBERT]

For anybody, including my self. I enabled socket file in InfluxDB and this require that the service has access to /var/run/influxdb.

Every time I update influxdb, I need to add the following 2 lines again to the influxdb.service under Ubuntu server:

[Service]
....

RuntimeDirectory=influxdb
RuntimeDirectoryMode=0755

Good hint, i did the following changes:

RuntimeDirectory=/var/lib/influxdb
RuntimeDirectoryMode=0755

[melroy89]

For anybody, including my self. I enabled socket file in InfluxDB and this require that the service has access to /var/run/influxdb.
Every time I update influxdb, I need to add the following 2 lines again to the influxdb.service under Ubuntu server:

[Service]
....

RuntimeDirectory=influxdb
RuntimeDirectoryMode=0755

Good hint, i did the following changes:

RuntimeDirectory=/var/lib/influxdb
RuntimeDirectoryMode=0755

That is not how RuntimeDirectory should be used.

Also the reason I created this issue is. Just changing this service is not really persistent. After each influxdb update, the file is overwritten and often put back to their defaults. Causing my instance to fail to start.