Logs

Runtime logs hold what a running service writes to standard output and standard error, and HTTP request logs record each request a web service answers. Each deployment also keeps its own build and deploy output, which Deployments covers, and each cron job run keeps its output under the job's Runs.

Read runtime logs

Open the project's Logs tab and pick a service. New lines appear as they arrive, and scrolling up stops following them until you choose Jump to latest. Jump to first error goes to the first line that looks like an error, and the Filter box narrows the view to lines that contain your text.

curl "https://api.dekka.cloud/v1/resources/$SERVICE_ID/logs?after=0&limit=500" \
  -H "Authorization: Bearer $DEKKA_API_KEY"

Each line has a seq, a timestamp, its stream and the line itself. To follow new lines, call again with after set to the next_after from the last response. A page holds up to 1,000 lines.

Read HTTP request logs

For a web service, choose HTTP requests next to the service picker. Each request shows its time, method, path, status, duration and response size. Narrow them with the Status menu, which takes a class such as 5xx, and Path starts with, such as /api.

curl "https://api.dekka.cloud/v1/resources/$SERVICE_ID/http-logs?status=5xx&path_prefix=/api" \
  -H "Authorization: Bearer $DEKKA_API_KEY"

The API also accepts an exact status such as status=404. Dekka never stores the client's IP address, only a salted hash of it as client_ip_hash. Workers and cron jobs don't serve HTTP, so they have no request logs, and asking for them answers 422 not_supported.

Search across the environment

Choose Search all to search the build and deploy output of every deployment, and the output of every cron job run, in the environment. Type a word such as timeout or a request ID, and narrow the results to the build, deploy or runtime stream. Results come newest first.

curl "https://api.dekka.cloud/v1/projects/$PROJECT_ID/logs/search?q=timeout&environment_id=$ENV_ID" \
  -H "Authorization: Bearer $DEKKA_API_KEY"

The search also takes resource_id, stream, since and until, and pages with the cursor from the previous response. Runtime logs of web services and workers aren't part of it, so read those from the service's own runtime logs.