Skip to content
GitHub
Start a 14-day trial

Logs

Open Logs and the last hour of your log lines is already on screen, newest first, with their volume charted above them. Narrow it with a sentence or a few filters, group it into patterns, or press Analyze and I’ll read the window and tell you what’s in it.

It lives under Exploration → Logs.

The Logs explorer over the last hour. The query bar reads Describe the logs you want — e.g. errors in checkout in the last hour, with Interpret, a 1h time range, a Live tail switch and Run beside it. Below are the Levels, Services, Environment and body contains filters, then 1,000 rows with Copy for agent, the Rows / Patterns and Chart / Table / Both toggles, and Insights. A bar chart of log volume per five minutes sits over a table with Time, Level, Service, Message, Span and Trace columns; the Insights rail on the right reads Press Analyze to interpret this log window.

Type what you want into the query bar — “errors in checkout in the last hour” — and press Interpret (or Enter). The filters it understood appear as pills and the query runs; a line under the bar says in plain words what ran. Set the window with the range control: a time in the question doesn’t change it. Ask in plain English covers how the bar reads your question.

The filters sit below the bar, always visible:

Filter What it does
Levels Pick one or more of TRACE, DEBUG, INFO, WARN, ERROR and FATAL.
Services Pick services; click a namespace heading to take all of its services at once.
Environment Choose an environment you’ve sent, or type one.
body contains… Keep the lines whose text contains what you type.

Each filter you set shows as a pill you can remove with ×. Logs re-runs the moment you change a level, a service or the time range; text you type into Environment or body contains… applies when you press Enter or leave the field. The range offers 15m, 1h, 3h, 24h and 7d — 1h the first time, then the last range you picked in any explorer — and Run fetches the window again. If a query comes back empty because of your filters, Logs names each filter that emptied it, with how many lines you’d see without it and a Remove filter button.

Rows lists the lines, 50 to a page: Time, Level, Service, Message, Span and Trace. Click a level or a service to filter by it. Expand a row and you get the whole body, the line’s timestamps and scope, its log attributes and its resource attributes; click any attribute value to add it as a filter.

The Span and Trace links open the request that wrote the line, with that span selected; the trace’s logs are in the panel below its waterfall. See Find in trace.

The chart counts lines by severity over the window. Click a bar, or drag across several, to narrow the query to that stretch of time; the window shows as a pill, and clicking the pill puts your previous range back. Chart / Table / Both picks what you see; Both is the default.

Patterns folds the same window into templates, with the variable parts — numbers, ids, timestamps — replaced by placeholders. Each pattern shows its Count, its Severity mix, when it was first and last Seen, and a few example Traces you can open. View logs switches back to Rows filtered to that pattern’s text, so you can read the individual lines behind a count.

Turn on Live tail and Rows refreshes every five seconds, so new lines arrive as your services write them. It follows Rows only; switching to Patterns stops it.

The Insights rail on the right is open by default. Press Analyze and I read the lines in the current window, then lay out what I found: how many lines I scanned, how many were errors, the severity mix, a summary, the Likely causes, the patterns behind them, and Recommended actions.

Copy for agent, beside the row count, copies the window — a summary, sample lines and my analysis once you’ve run it — as text for your own coding agent.

The filters ride in the page address, along with Rows or Patterns, so a refresh, the Back button or a link you send a teammate opens the same query.