RealUptime Errors
Uptime says 100%. Your checkout has thrown 1,204 times.
RealUptime Errors is error tracking that never lies about a count. It groups your exceptions into issues, tracks them across releases, and alerts your team the moment one is new or regresses. Signing up for Errors on its own isn't available yet.
The number looks precise. It is what survived sampling and rate limits; the 1,204 events that did not are a footnote, or nothing at all.
TypeError in checkout.ts
Example- New in v2.4.1Last seen 2 minutes ago
- 312 occurrences shownsince 09:14
- 1,204 events not ingested1,100 over quota, 104 rate-limited
A dropped event gets counted and shown, on the issue, the project, and the period totals, never silently absorbed into a smaller number. A count we publish is a count we stand behind, or it carries its asterisk in plain words.
Grouped, not flooded
Every raw exception collapses into one issue.
TypeError in checkout.ts:142
Illustrative example. The same fingerprint groups every occurrence into one issue, and known PII fields are scrubbed before an event is stored.
The issue screen
The count, and the events that never made it in, in one frame.
Most trackers put the number you can act on in the headline and the number they threw away in a settings page, if anywhere. Both belong on the issue, because a count you cannot check is not a measurement, it is a claim.
TypeError in checkout.ts
Cannot read properties of undefined (reading 'rate')
- 1,100 over quota
- 104 rate limited
Both numbers live on the issue, the project, and the period totals. A dropped event is counted where you can look, not absorbed into a smaller number that reads as precise.
First seen 08:41, last seen 2 minutes ago. Same fingerprint, one issue, however many times it throws.
Simulated issue on an invented codebase, not a captured exception from any project.
What comes with it
Everything the product ships with.
Grouped by fingerprint
Exceptions collapse into issues, not a firehose of duplicates, with search, snooze, and saved searches.
Releases
See which release introduced an issue and which one resolved it.
Honest drop accounting
When the event quota gates, the dropped count is shown on your dashboard, never silently discarded.
Optional overage
Off unless you turn it on, capped by a budget you set.
How it works
From setup to signal in three steps.
01
Install the SDK
Add the client to your app and errors start arriving grouped into issues within minutes. JS/Node and Python, one wire contract.
02
Triage by issue, not by event
Fingerprints collapse duplicates; releases show where an issue started and where it stopped.
03
Trust the counts
Every dropped event is counted where you can see it, so a quiet dashboard means quiet code, not a silent quota.
RealUptime Errors
Error tracking that never lies about a count.
Errors is a separate product with its own plans. Drop in the JS/Node or Python SDK and get grouped issues, releases, and alerts, with PII scrubbed before it leaves your process. If we ever drop an event, at quota or anywhere else, the count and the reason are on your dashboard.
Errors Free
$0
No card required
For a side project or the first app you want honest error counts on.
Real error tracking for a small production app, free forever.
- 10,000 events a month
- 14 days of retention
- JS/Node and Python SDKs, PII scrubbed by default
- Grouping, releases, and alerts on your existing channels
- Every dropped event counted where you can see it
Errors Growth
$23.99/mo
or $239.90/yr, 12 months for the price of 10
For a team whose app has real traffic and real on-call.
Ten times the events and a quarter of retention, on top of everything in Free.
- 100,000 events a month
- 90 days of retention
- Everything in Errors Free
Errors Scale
$58.99/mo
or $589.90/yr, 12 months for the price of 10
For services busy enough that error volume is itself a signal.
Three quarters of a million events and half a year of retention.
- 750,000 events a month
- 180 days of retention
- Everything in Errors Growth
Four products, one incident
Three products hold a piece of the answer. The fourth one publishes it.
Every product here is worth running on its own. Run more than one and they stop being separate graphs: signals that share a time window get joined into a single window with a verdict, so the question an incident actually asks, is this us or is this upstream, has an answer instead of three tabs.
Your infrastructure looks healthy
Regional checks and the server agent's own CPU, memory, and disk thresholds, so an incident can be ruled IN or OUT on your side.
- RealUptime ErrorsYou are here
One issue's rate just jumped
The issue that started throwing, when it started, and which release it arrived in, counted honestly including what was dropped.
A vendor you depend on is degraded
Our own independent probes on the services you named as dependencies, not the vendor's own status feed.
Probable upstream incident. Your infrastructure appears healthy.
The window names every signal it used and every one it could not get, and it says probable, never confirmed, because a joined time window is evidence and not proof.
The incident your customers read, drafted from what was actually measured, with the affected regions named.
RealUptime Complete is what paying for more than one of these is called, and the bundle discount is on the pricing page. The discount is the small part. The reason to run them together is the sentence above.
See it live
This site reports its own exceptions here.
Watching servers, not code? See RealUptime Monitor. Heartbeat (cron) monitoring for jobs that silently stop running is part of Monitor, a separate billable product, not a bundled Errors feature.