Analytics cookies only (Google Analytics, PostHog). No ad networks, no cross-site tracking. See our privacy policy.
A search engine built for logs and traces, with full-text search and aggregation queries, an Elasticsearch-compatible API that works with Elasticsearch and OpenSearch clients, and native OpenTelemetry and Jaeger support. It indexes schemaless or strict-schema JSON, can keep its index files on cloud object storage such as Amazon S3, Azure Blob Storage or Google Cloud Storage, and separates compute from storage with stateless indexers and searchers. A single container runs a whole node with the web UI and REST API on port 7280.
Runs on a VPS from $96/mo (Linode Linode 16 GB, 16 GB RAM) · prices checked Nov 2026
Fits home hardware (21 MB measured idle, RAM only): Raspberry Pi 4 (4 GB) (arm64 untested), Raspberry Pi 5 (8 GB) (arm64 untested), Repurposed laptop (8 GB), Intel N100 mini PC (16 GB), AMD Ryzen mini PC (32 GB).
Pick Quickwit when you want to search and aggregate logs or traces in bulk and keep the index cheap, rather than add search to a website. It is a search engine written in Rust under Apache-2.0, with full-text and aggregation queries, an Elasticsearch-compatible API, native OpenTelemetry and Jaeger support, and index files that can sit on S3, Azure Blob Storage or Google Cloud Storage with stateless indexers and searchers. A single container is a whole node with the REST API on port 7280. Our snippet runs `quickwit/quickwit:0.9.3` (released 2026-10-22) as the quickstart's `quickwit run`, with a named volume for `/quickwit/qwdata` and a restart policy, and then runs the quickstart's steps on two sample documents: create an index, ingest, and search for one hit. The quickstart publishes the port on `127.0.0.1`; ours publishes it on all interfaces. Upstream's cluster sizing guide says not to use indexer instances with less than 8 GB of RAM or searcher instances with less than 4 GB, and a `quickwit run` node does both, so our floor is 8192 MB even though the empty node idled far lower. In the catalog, Meilisearch and Typesense are typo-tolerant search for a site or app, Manticore is a SQL-queried engine, OpenSearch is the Elasticsearch fork with Dashboards, and SearXNG is a metasearch frontend; Quickwit is the one that stores its index on object storage and targets log and trace volumes.
Paid link — we earn a commission if you shop through it.
Run these commands on your server:
# Quickwit — adapted from the Docker instructions in the upstream quickstart (quickwit/quickwit run, port 7280, data in a volume), image pinned to quickwit/quickwit:0.9.3
# Changed from upstream: a Compose file with a named volume and restart policy, port 7280 published on all interfaces instead of 127.0.0.1, then the quickstart's create-index, ingest and search steps run through the REST API on a small sample so the install proves itself
mkdir quickwit && cd quickwit
cat > docker-compose.yml <<'EOF'
services:
quickwit:
image: quickwit/quickwit:0.9.3
command: ["run"]
restart: unless-stopped
ports:
- "7280:7280"
volumes:
- qwdata:/quickwit/qwdata
volumes:
qwdata:
EOF
docker compose up -d
timeout 300 bash -c 'until curl -fsS -o /dev/null http://localhost:7280/api/v1/version; do sleep 3; done'
cat > notes-index.yaml <<'EOF'
version: 0.8
index_id: notes
doc_mapping:
field_mappings:
- name: title
type: text
- name: body
type: text
indexing_settings:
commit_timeout_secs: 10
search_settings:
default_search_fields: [title, body]
EOF
curl -fsS -X POST http://localhost:7280/api/v1/indexes -H 'content-type: application/yaml' --data-binary @notes-index.yaml -o /dev/null
printf '%s\n' '{"title":"Lantern","body":"A lantern lights the harbour at night"}' '{"title":"Compass","body":"A compass points north"}' > notes.ndjson
curl -fsS -X POST 'http://localhost:7280/api/v1/notes/ingest?commit=force' --data-binary @notes.ndjson
curl -fsS 'http://localhost:7280/api/v1/notes/search?query=lantern' | tee search.json
grep -q '"num_hits": 1' search.json
# → http://SERVER_IP:7280 (built-in UI and REST API; there is no authentication, so keep the port closed to the internet or put a proxy in front)Quickwit's README, configuration docs and changelog flag these before you rely on it:
Quickwit used 21 MB of RAM at idle on GCP e2-standard-2 · Ubuntu 26.04 LTS · Docker 29.9.3 (first boot, measured Nov 2026). We list a minimum of 8192 MB.
The Quickwit install snippet is marked verified on Ubuntu 26.04 (Nov 2026).