Free server tool / Performance diagnostics

A clearer picture of what slows you down.

Bring a Spark report. Read the signals, follow the expensive work, and leave with a useful next step.

TPS & tick timeMemory signalsPlugin attribution

01 / Bring your capture

Turn a profile into a next step.

Analyze tick health, memory pressure, and sampled add-on costs. No account required.

Or upload raw JSON

Analysis runs in your browser. Links fetch report data directly from Spark; uploaded JSON stays on your device. Maximum 12 MB. No report is saved by this tool.

Need a fresh capture? Pick the problem.

Run during the busy period you want to investigate.

/spark profiler start --timeout 180

If Spark already has a background profile running, use /spark profiler open to view it. Spark command guide

Your server has a story. Start with a capture.

Paste a profiler or health-report link above. A sample is available if you want to see how the findings and evidence fit together.

Reading your result

From a signal to a useful next step.

01 / Read the verdict

Start with tick rate and the time budget. The verdict describes the metrics available in this capture; missing metrics or a quiet session cannot rule out a problem.

02 / Follow the evidence

Each finding pairs an observed value with a next step. Work through severe findings first, then use the timeline to see where unusually long ticks appeared.

03 / Inspect the work

Choose a captured thread and compare add-on or method costs. A large share shows where to look in Spark’s original viewer, not who to blame automatically.

Start with good evidence

Capture the problem while it happens.

01 / Start a profile

Run Spark while the server is experiencing the workload you want to understand.

/spark profiler start --timeout 180

02 / Bring the result

When the profile finishes, copy the viewer link into the tool above. A profiler report can include call-tree attribution; a health report provides metrics without those costs.

03 / Compare one change

Inspect the evidence, change one setting, and capture again under similar load. Keep the original profile available when discussing a finding with a plugin author.

Commands and report format: Spark documentation. Read the original viewer guide for deeper call-tree inspection.

A little help before you launch

What this tool can tell you.

Understand the evidence, the gaps and the next step.

Let’s talk it through.

Have a question of your own? Our team is here to help.

Ask our team (opens in a new tab)Or explore the guides

5 questions

01Are my reports uploaded to EnderHost?

No. A pasted link fetches data directly from Spark’s public endpoint. JSON files are processed locally in your browser and are not uploaded. The tool does not persist report data. Sharing a report link gives its recipient access to the report on Spark.

02Does a large plugin percentage prove it is the problem?

No. The analyzer attributes each branch once to the first mapped add-on, including downstream work. Percentages are sampled time on the named thread, not CPU utilization. They help you choose where to inspect, but require context and a controlled follow-up capture.

03Why are some fields blank?

Report types and Spark versions include different fields. A missing metric is shown as unavailable. Allocation profiles do not receive execution-time attribution, and reports without server tick statistics cannot establish server tick health.

04What if my report has expired?

Capture a new profile, preferably during the same problem. You can also upload a saved raw JSON export. Reports larger than 12 MB should be recaptured over a shorter interval.

05Should I upgrade my memory?

Inspect heap behavior and garbage collection first. A single high heap value can fall after collection. For initial resource planning, try the Minecraft RAM calculator.

Still have a question? Talk to our team

More free server tools