Monitoring

[gazwald]

#108

[darkshade9]

Free third party accounts to check availability like Checkly?
Prometheus for metrics gathering internally?

[gazwald]

Initially I thinking of exporting metrics to CloudWatch and generating alarms from those. Prometheus would be a good option as well. Depends on the pricing model in AWS. I figure the interface would be similar to how Storage is implemented so it's relatively simple to swap between services.

Synthetic tests (e.g. something sending a ping and expecting an ack) was another route I was planning with the client CLI. Probably spawned from a scheduled AWS Lambda or something.

Tracing would be awesome (X-Ray/Jaeger) but definitely something for later on. Need to divide all the options up into the minimum required for now, and required for if enough people start relying on the service.

[darkshade9]

My day job is as an SRE/DevOps, so let me know what you think would work out best. Running an internal Prometheus container could fit the bill to keep tabs on utilization, and external synthetics for user-facing uptime/performance, integrate them into an alerting service for specific triggers, etc.