Watch it happen, then keep it

Tail an HTTP(S) endpoint into a bounded buffer and get the same filters, search and level toggles you'd have on a static file. When something interesting lands, save the buffer and carry on analysing it after the stream ends.

Lives in the nav rail — the second icon down

412 lines/min
Fig. 01edge-01 · streaming
Search…
LLogier
https://logs.internal/edge-01/streamAuthorization: Bearer ••••
Pause
buffer 5,000 lines · 412/minalert sent · 14:22
INFOWARNERROR
Paused 14:21:04 – 14:21:52 · gap recorded
TimeStatusHostMessage
14:22:08INFOedge-01health ok — 412 rps
14:22:09WARNedge-01connection pool at 87% capacity
14:22:09ERRORedge-01upstream timeout after 5000ms
14:22:10ERRORedge-01upstream timeout after 5000ms
14:22:10ERRORedge-01retry queue exceeded configured limit
14:22:11WARNedge-01connection pool at 94% capacity
14:22:11ERRORedge-01upstream timeout after 5000ms
14:22:12INFOedge-01shedding load — 38 requests rejected
14:22:12ERRORedge-01upstream timeout after 5000ms
14:22:13ERRORedge-01connection reset by peer
14:22:13WARNedge-01circuit breaker half-open
The pause control is on screen because pausing is what creates a real gap. The break in the chart at the fourth bucket is a recorded interval, not missing data drawn as zero.

Five things it does while connected.

  1. 01

    Connect

    Point it at an HTTP(S) endpoint whose body is a stream of text. Custom request headers are set as JSON, so an Authorization bearer token or an internal API key goes in before you connect.

  2. 02

    Parse as it lands

    Bytes are read chunk by chunk and split into lines, then run through the same detection as a static file. Filters, search and level toggles behave identically — there is no separate live-only mode to learn.

  3. 03

    Pause honestly

    Pausing records a real interval, and the chart draws that as a break rather than closing the line over it. You can always see that you looked away, and for how long.

  4. 04

    Alert once, not per line

    Incoming batches are checked against your severity triggers and fire a single summarising email, subject to a cooldown — sent through the main process, because the email API sends no CORS headers.

  5. 05

    Keep what you saw

    “Save to editor” turns the current buffer into a snapshot file you can keep scrolling, filtering and searching after the stream ends.

what it iswhat it isn't

Four things it is not.

This is the feature people most often assume is something else. Better to find out here than after paying — and if any of these four is what you actually need, Logier is the wrong tool and we’d rather say so.

  • Not a file tailer

    It doesn't watch a path on disk. The source has to answer over HTTP or HTTPS with a streaming text body.

  • Not an SSH agent

    There's no shell, no key handling and no jump host. If the only way to the log is SSH, put something in front of it that speaks HTTP.

  • Not a 24/7 monitor

    It's scoped to an open window on your machine. Close it and the stream ends. For unattended production monitoring you want a real observability stack.

  • Not unbounded

    The buffer holds 5,000 lines by default. Past that, the oldest drop off as new ones arrive, so a long session doesn't grow memory forever.

faq

Common questions

keep reading

Where this fits

  1. 01

    Timeline view

    The same chart, redrawn as entries land — and the place a recorded pause shows up as a real break.

  2. 02

    Auto-format detection

    A stream gets the same parsing as a file, so a remote endpoint doesn't need its format declared either.

For the logs you can't SSH into.

An endpoint, a bearer token, and the same filters you'd use on a file. Save the buffer when something interesting shows up.