Support reloading the configuration file with SIGHUP

[jsternberg]

Feature Request

Support the standard reloading of a configuration file with SIGHUP for anything that would be reasonable.

Proposal: Support reloading the configuration file with SIGHUP.

Current behavior: You must restart InfluxDB to load new configuration options.

Desired behavior: Configuration values that are easy to change (such as log levels or query limits) without requiring a potentially long restart.

Use case: It allows a DevOps person to be able to update certain settings when tuning the database without downtime.

[candlerb]

Relates to influxdata/telegraf#69

[Darkentik]

Relates to influxdata/telegraf#69

You missunderstand him. He not wanted to reload the telegraf instead of the database influxdb.
this is a feature we need too. I always have to restart the complete influxdb service and this ends up with some downtimes of the influxdb.
Or does this exists already? I didn´t found something in the docs.

[candlerb]

By "relates" I meant "similar feature requested in a different component of the TICK stack"

[galindro]

I need this feature too because I'm using letsencrypt to generate influxdb certificates and a reload configuration could be used to load the new certificate, avoiding database downtime.

[Genda1ph]

I'm bewildered how this still hasn't been implemented.
Is there a different way for online configuration change?

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[Genda1ph]

Not having any activity doesn't mean it's not important.

[phgogo]

This is a feature i expect in an production-ready application!

[m82labs]

Putting a vote in for this as well.

[ruairinewman]

We would definitely like to see this implemented also. With production cluster nodes taking as long as 4 hours to complete a restart due to the volume of data we hold, this would avoid many problems.

[m82labs]

We would definitely like to see this implemented also. With production cluster nodes taking as long as 4 hours to complete a restart due to the volume of data we hold, this would avoid many problems.

Just curious, but how much data do you have? I have about 1TB and a service restart takes about 5 minutes. Interested to know when I might get to 4 hour territory.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[Genda1ph]

bump

[lhanson]

Bumping again because it seems incredible that I can't update configuration without downtime.

[foobar]

any response from maintainers?

[vipulsharma21]

bump

[anandpathak]

bumping! this is a must-have feature

[fmr-ap]

bump

[suhlig]

I'd like to have this.

[MonilCereBulb]

+1

[TiagoMHCSantana]

+1

[Guzya]

+1

[sobos-office]

+1

[hardincm2]

+1

[Technoguyfication]

This is a big problem for me, I have to automatically restart influx whenever my SSL cert renews. I'm able to reload nginx with new certs each time, so why can't I reload influx?

[danielelopez1]

+1

[sdrsdr]

+1

[dbivolaru]

+1

[emilbasv3]

+1 This is a must have for a production ready product.

[joelsdc]

I have to admit I was surprised to see this issue still open from 2016, I would have expected to have a way to reload SSL certs since day 1.

As a workaround, we've added our load balancers (Kong) in the path to reach influxdb, but it would be nice to not have to proxy requests just for SSL.

[Genda1ph]

Thank you for reminding me to add this date to my anniversary calendar.
I ended up running Influx in Docker and passing requests through Traefik. A shame features essential to running huge databases are being ignored, but oh well.

[viveklanjekar]

is everyone still waiting for this basic feature in 2022?

[Genda1ph]

Sixth anniversary...

[abhi-jha]

So, what happened?

[Genda1ph]

checks notes
exactly... Nothing.

[Darkentik]

+1
Offtopic (sry for that, but i can't hold back my anger):
This is sadly not the only poor (missing) feature. Have you messed up with backup and restore?
It is a disaster, because they change the bucket ID and this brokes everything while using tokens with permissions for this bucket ID...we have figured out to get the same bucket ID as before but holy moly...seriously? This can't be the right way to handle this.

We ask us how influxdb2 can be used from so much users and then there are such holes in functionality.

[enspritz]

I'm going to take gamble and assume that I speak for everyone here in asking for a response from our friends on the Influx team.
The ability to hot-reload configuration is becoming table stakes, for TLS, for avoiding downtime, and myriad other reasons.
Hope to see this delivered soon ^_^

[Genda1ph]

I want to congratulate everyone with 7th anniversary of this bug!
Here's to meeting all of you in a year!

PS: labels need to be updated.

[SuperCoolAlan]

Hoping to see this feature soon! I am sending SIGHUP to influxdb2 when TLS keys are rotated, but influxdb is not exiting gracefully.

[sergejhorvat]

+1