quickwit
quickwit copied to clipboard
Cloud-native search engine for observability. An open-source alternative to Datadog, Elasticsearch, Loki, and Tempo.
Currently, scaling indexers and searchers is slow as we use the `OrderedReady` pod management policy (it's the default). It seems very interesting to use the parallel policy to start all...
We could probably add support for VRL in the `DocProcessor` for instance.
Fixed logic for cleaning up local split store cache
It is currently not possible to use a Sasl connection with a Kafka source. Currently you would get the following error message with a SASL configuration: ``` Command failed: Failed...
A Quickwit node is meant to support thousands of indexes, while maintaining a decent throughput, and small memory footprint. Currently the Ingest API works by using RocksDB as our storage....
1. The parsing API prevents us from validating custom input formats submitted by users 2. The lib does not seem maintained and has a bunch of lagging dependencies. We can...
In Quickwit, we have some special objects with known names. - SOURCE_FIELD_NAME: `_source` - DYNAMIC_FIELD_NAME: `_dynamic` - CLI_INGEST_SOURCE_ID: `.cli-ingest-source` - INGEST_API_SOURCE_ID: `.ingest-api` Let's hamonise all this to use the `_`...
We can take inspiration from meilisearch CI: https://github.com/meilisearch/meilisearch/pull/2837/files
Position of the MSB: - 28-33 -> seconds - 38-43 -> millis - 48-53 -> micros - 58-63 -> nanos TODO - [ ] implement MSB-based UNIX timestamp parsing -...
Inspect whether the `merge_enabled` argument in this function makes snese. ``` async fn spawn_merge_pipeline( &mut self, ctx: &ActorContext, index_id: String, merge_enabled: bool, ) -> Result { let pipeline_id = IndexingPipelineId...