Create and manage pre-allocated API keys in bulk

[mloskot]

I'd like to bulk-create API keys with Ingest permission, title and specific properties to be authenticate numerous apps for Seq.
I'd like to utilise the feature of A pre-allocated API key token, so I can stick pre-generated API keys in a config file.

After learning Seq over the last weeks, I have some ideas, which I will be happy to share once I have them tested, but I'd like to validate some basic of my understanding first.

Questions

1. First and foremost, is it reasonable to use the init scripts to mass-create API keys?
2. Can I use seqsrv command in any way to mass-feed the Seq with API keys or shall I stick to the seqcli and its commands?
3. Can I use seqcli apikey create to update pre-existing API keys in place? In other words, if my initial input config contains 5 API keys and I add 5 more later, and re-run the script running seqcli apikey create, then do I have to take care of NOT re-running it for the 5 keys created in the initial run and only run the command for the new incoming keys? Or will Seq cleverly just update the pre-existing API keys in place, for example, adding new properties?

I'll be thankful for any insights.

Alternative to init scripts

I am using Seq in Kubernetes.

I understand the Init Containers is not an option due to the chicken-and-egg issue. Unless, I run Seq instance in an init container with same storage mounted from the same PV that will be mounted and used by the main container. There surely must be more caveats.

Some Seq users seem to use Kubernetes Job to run some administrative actions.

The init scripts seem to be clean and native solution though.

[liammclennan]

Hi @mloskot

To answer your questions:

1. API keys are created via the API. Seq is not running when init scripts run so the API is not available.
2. API keys are created via the API, so not seqsrv.
3. API keys are updated via https://docs.datalust.co/docs/command-line-client#apikey-update.

[mloskot]

Hi @liammclennan,

Thanks for those clarifications.

Somehow I overlooked the apikey update command that can take JSON.
(It would be useful if there was 'symmetric' feature of apikey create accepting JSON too - from the docs it looks it doesn't.)

It looks like the sensible approach will be to run the bulk create-or-update API keys step using seqcli indeed and as Kubernetes Job, and as soon the Seq /health responds with status healthy.