How to monitor Hasura performance ?

[Aarbel]

Hi Hasura team and community.

We are using Sentry.io performance monitoring on both front-end (React) and Back-end (Apollo).

Problem: Hasura has no integrations with Sentry and it's like a blackbox hard to monitor, especially about performance metrics.

What are the best tools / methodologies to monitor Hasura internal performance ?

Thanks a lot

[adas98012]

Hi,
Thanks for reaching out. Hasura currently supports the following external tools/services to monitor performance logs/metrics (Reference Doc: https://hasura.io/docs/latest/graphql/cloud/metrics/integrations/index/)

Thanks,

[kokroo]

Making something as basic as observability a paid feature, on software that has contributions from people around the world is just laughable.

Hasura is slowly nerfing the self-hosted version, even though there are numerous code contributions done by everyday users for free.

[Aarbel]

@adas98012

Would we have some ways with Hasura V3 to include sentry to monitoring what happens inside Hasura self hoested ?

Thanks a lot !

[Aarbel]

@adas98012 do you have any news about performance monitoring without having to pay hasura ? (self hosted)

[Aarbel]

@adas98012 i just saw that this performance tracing can begin with Hasura Query Tags, but this is also an only entreprise plan.
Looks like you are locking all the basic features to force people to pay, even if they use open source self hosted.

Can you please unlock at least this feature (Hasura Query Tags) for self hosted (non entreprise plans) ?

Cf #10224

[Edmartt]

Stop following this project. They are just taking advantage of people for making everything private with their dual license like the new Redis.

[hgiasac]

FYI some of Hasura cloud monitoring services collect logs and ingest metrics via ETL tools. That means it is also possible with Hasura OSS as well. Most observability services such as Datadog, HoneyComb, and Google Cloud Logging support transforming metrics from logs.

If you want to export Prometheus, or OpenTelemetry metrics from logs, vector and OpenTelemetry Collector can help you. There are many alternatives there.

See GraphQL Engine logs docs for log structure examples. They are in JSON format and easy to be parsed.

If you want a native OpenTelemetry exporter, let's try the Hasura v3 ecosystem. The Hasura v3 OSS version and data connectors natively support OpenTelemetry. You can fork, contribute, or customize your connectors as you desire with any language. Rust is also easier to understand than Haskell.

https://hasura.io/docs/3.0/observability/traces