Yes—there are credible alternatives now, including several that are dramatically easier to self-host than Sentry.
My shortlist
| Product | Best for | Self-hosting footprint | License | Main limitation |
|---|---|---|---|---|
| GlitchTip | Sentry-like error tracking | Very light | MIT | Less sophisticated than Sentry |
| Bugsink | Extremely focused error tracking | Very light | PolyForm Shield | Source-available, not OSI open source |
| HyperDX | Logs, traces, errors and session replay | Moderate | MIT | ClickHouse-based; not as light as GlitchTip |
| OpenObserve | Graylog replacement and unified telemetry | Light–moderate | AGPLv3 | Error workflow is less Sentry-like |
| Uptrace | Application performance and distributed tracing | Moderate | AGPLv3 | Error triage is secondary to APM |
| SigNoz | Full AppSignal/Datadog-style observability | Moderate–heavy | Apache 2.0 core | More infrastructure than the focused options |
1. GlitchTip — my first choice for replacing Sentry
GlitchTip is probably the closest match to my priorities:
- Compatible with existing Sentry SDKs and DSNs.
- Error grouping, stack traces, releases, source maps and alerts.
- Also has basic performance monitoring, uptime checks and logs.
- Runs as one application service plus PostgreSQL; Valkey/Redis is optional.
- Officially recommends only 512 MB RAM, with 256 MB possible for small installations.
- Straightforward Docker Compose upgrades.
- Permissive MIT license.
It deliberately chooses simplicity over Sentry’s enormous feature set. That is likely a feature in my case, not a drawback.
The tradeoff is that its performance monitoring, search, release intelligence and issue workflow are less sophisticated than contemporary Sentry. But if your primary requirement is “tell developers what failed, where, and with what context,” this is the strongest candidate.
GlitchTip overview · self-hosting requirements · source and license
2. Bugsink — even more focused and operationally simple
Bugsink exists almost precisely because its creator found self-hosted Sentry too difficult.
It offers:
- Sentry SDK compatibility—migration is generally just changing the DSN.
- A single application container rather than dozens of services.
- Error grouping, releases, alerts, retention rules and local-variable context.
- Approximately 2 GB RAM recommended.
- Optional paid support specifically for self-hosted deployments.
Its important caveat is licensing: Bugsink is source-available under PolyForm Shield, not open source under the OSI definition. You may inspect, modify and run it internally, but cannot offer a competing hosted service.
It also intentionally handles only error events—not traces, metrics or general-purpose logs. That focus is why it remains simple.
Bugsink documentation · Sentry SDK compatibility · license comparison
3. HyperDX — best integrated developer experience
HyperDX is the option I would investigate if you want something broader than GlitchTip without assembling the conventional Grafana/Loki/Tempo/Prometheus collection.
It brings together:
- Logs
- Traces and APM
- Metrics
- Exceptions
- Browser session replay
- Dashboards and alerts
- OpenTelemetry ingestion
It is MIT-licensed and now part of ClickHouse’s ClickStack. For evaluation, an all-in-one container includes HyperDX, ClickHouse, an OpenTelemetry collector and MongoDB. The documented test footprint is at least 4 GB RAM and two CPU cores.
This is considerably simpler than Sentry operationally, although it is still a real observability stack. Its strongest feature is correlation: jump from an exception or slow request to its trace and related logs.
HyperDX source, features and deployment
4. OpenObserve — strongest Graylog replacement
If centralized logs are the biggest need, OpenObserve deserves serious consideration.
It provides:
- Full-text and SQL log search
- Logs, metrics and traces
- Dashboards, alerts and ingestion pipelines
- OpenTelemetry, Fluent Bit, Vector and Prometheus support
- S3-compatible object storage for economical retention
- A single-binary deployment option
The community edition is AGPLv3. Architecturally, it is likely the simplest unified backend in this list. It is especially attractive if you have lots of logs and want to avoid running Elasticsearch or ClickHouse yourself.
Its weakness is that application exceptions are treated more as telemetry records than as a polished Sentry-style issue inbox. It can show and alert on errors, but GlitchTip is better for developer-oriented error grouping and triage.
OpenObserve documentation · source and licensing
5. Uptrace — compact, capable APM
Uptrace is a good AppSignal alternative when performance monitoring matters more than session replay or Sentry-style issue management.
It includes:
- OpenTelemetry traces, logs and metrics
- Service graphs
- Database and request performance
- Automatic dashboards
- Alerting
- Prometheus/Grafana compatibility
The server is AGPLv3. Its principal backend is ClickHouse, with PostgreSQL used for metadata. That makes it more involved than GlitchTip or standalone OpenObserve, but substantially less sprawling than Sentry.
Uptrace source and architecture
6. SigNoz — closest to a complete AppSignal/Datadog replacement
SigNoz offers the most complete conventional observability package here:
- Logs
- Metrics
- Distributed tracing
- APM
- Infrastructure monitoring
- Dashboards and alerting
- Native OpenTelemetry support
The core repository uses Apache 2.0, with some enterprise functionality separated. Recent deployment tooling can generate and operate Docker Compose or systemd installations from a single configuration.
I would not describe it as lightweight in the same sense as GlitchTip. It is still based around an OpenTelemetry collector and ClickHouse, and its total operational surface is appreciably larger. But it is much more coherent than assembling six unrelated CNCF components.
SigNoz repository · current deployment tool
What I might choose
For the lowest-maintenance practical setup:
- GlitchTip for exceptions and developer error triage.
- OpenObserve for logs, traces and possibly metrics.
- Send everything through OpenTelemetry Collector or Vector, so the applications are not permanently tied to either backend.
That gives two comparatively simple systems, each doing what it is good at. It will usually be easier to understand and recover than one ambitious all-in-one platform.
For a single UI:
- HyperDX — best developer troubleshooting experience.
- OpenObserve — simplest architecture and strongest log orientation.
- SigNoz — broadest mature APM feature set, but more machinery.
If error tracking is overwhelmingly the priority:
- GlitchTip if true open source matters.
- Bugsink if operational simplicity and paid self-hosted support matter more than OSI licensing.
GlitchTip versus HyperDX: GlitchTip for a clean Sentry replacement, HyperDX to replace Sentry, Graylog and part of AppSignal with one system.