New: Replay
The ticket says what failed. Replay shows how it got there.
Open a ticket and watch the logs of all your services arrive in order, up to the moment of the error. When the cause started somewhere else, you see it.
No card. One project, 5M events a month, 7 days of replay.
highorders · Example ticket
Database connection pool exhausted in orders
Requests to orders fail while waiting 30 s for a database connection. Probable cause: more concurrent queries than the pool can serve.
14:22:0010:00 before ticket
| orders | 188 lines | 4 warnings | 4 errors |
| billing | 124 lines | 2 warnings | 2 errors |
| gateway | 1,442 lines | 2 warnings | 0 errors |
| auth | 120 lines | 0 warnings | 0 errors |
| mailer | 60 lines | 0 warnings | 0 errors |
| search | 60 lines | 0 warnings | 0 errors |
All services, lines up to the cursor
No lines up to this moment.
Example data. The player is the one in the product; the notes beside it are ours, for this page. In the product you see only the lines, and the conclusion is yours.
Why
An error's cause rarely sits next to the error.
With the ticket alone
- What failed, how often, the stack trace and a probable cause.
- The 20 lines before the error and the 10 after, from one service.
- At 100 logs a second, those 20 lines cover 0.2 seconds.
With Replay
- Everything the project logged in the 15 minutes before, and back up to 24 hours.
- Every service on one time axis, so a cause in another service is in view.
- Warnings and errors marked on the timeline, so you jump to them instead of scrolling.
Investigate
The questions you ask after an incident.
What happened just before?
Play the logs that led to the ticket at 1x, 10x or 60x, and go back up to 24 hours. Playback stops at the ticket, and up to 5 minutes after it can be read.
Where did it start?
Each service has a row, with its volume and its worst level. Click a service to read only its lines.
Has it happened before?
An error group's history shows occurrences, tickets and a weekday and hour heatmap. Each ticket opens its own replay.
Did a release change it?
The group's history marks the releases that reported the error, next to its occurrences.
The evidence
Raw lines. Nothing filtered, nothing interpreted.
No AI in the replay
No generated causes or summaries over the lines. It shows the evidence and leaves the conclusion to you.
Redacted before it is stored
Lines go through the same redaction as everything else, so emails, tokens and IP addresses appear masked. Each message is cut at 2 KiB.
A gap never looks like silence
If lines could not be stored, the replay says how many and when. Time before a project's history started is marked as such.
You can turn it off
An admin can turn off the log history of any project. Retention follows your plan, and lines past it are deleted.
Lines are written every 5 minutes on Free and every 2 minutes on paid plans, so the last minutes of a new ticket arrive shortly after it. Replay starts from a ticket: it is not a log search. trace and debug lines appear only if you send them.
Read the Replay guideGet started
From sign-up to your first replay.
Step 1
Create your account
Free, without a card.
Step 2
Send your logs
Two lines in your deploy with the Node.js, Python or Go SDK, or an OpenTelemetry Collector.
Step 3
Open a ticket and press Replay
Replay needs no setup: it reads the lines stored for the project.
Node.js
npm i @dlogify/node LOGIFY_API_KEY=lgf_live_… node --import @dlogify/node/register server.js
Python
pip install dlogify LOGIFY_API_KEY=lgf_live_… logify run -- python app.py
Plans
Replay is on every plan. Only the days change.
Free
7 days
of replay and group history
Pro
30 days
of replay and group history
Team
90 days
of replay and group history