This change integrates async-profiler 4.1 with two profiling modes for QuestDB performance analysis.

Usage

1. Continuous Profiling (start with agent)
   Start QuestDB with profiling enabled from startup:

  $ questdb.sh start -p

Profile files will be saved to ~/.questdb/profiles/ (or your configured $QDB_ROOT/profiles).

2. Attach to Running Instance
   Attach profiler to an already running QuestDB instance:

  # Start a CPU profile for 30 seconds and save as HTML flame graph
  $ questdb.sh profile -- -e cpu -d 30 -f /tmp/profile.html

  # Start allocation profiling and save as JFR
  $ questdb.sh profile -- -e alloc -f /tmp/alloc.jfr

Custom Profiler Parameters

Override default profiling settings:

# Continuous profiling with custom parameters
$ questdb.sh start -p -- start,event=alloc,file=/tmp/custom.jfr,interval=1ms

# Profile specific database instance
$ questdb.sh profile -t mydb -- -e cpu -d 60 -f /tmp/mydb-profile.html

Analyzing Profiles

Generate CPU flame graph heatmaps from the JFR files:

$ jfrconv --cpu -o heatmap ~/.questdb/profiles/profile-1.jfr heatmap.html

Note: jfrconv is part of the async-profiler project and is included in a QuestDB Linux x86-64 distribution under the lib/ directory.

Features

• Two profiling modes: continuous (agent-based) and on-demand (attach to running instance)
• Continuous profiling with configurable loop duration (default: 30 minutes)
• JFR format output with timestamped files in $QDB_ROOT/profiles/
• Configurable profiling parameters (event type, sampling interval, etc.)
• Support for custom agent parameters via -- delimiter
• Automatic profile directory creation
• Robust error handling and validation
• Fails fast if profiler library is missing when -p is specified

Currently implemented for the Linux x86-64 distribution only. The libasyncProfiler.so library and asprof/jfrconv binaries are included in the distribution package under the lib/ directory. For other platforms you download the async profile from https://github.com/async-profiler/async-profiler/releases and copy asprof and libasyncProfiler.so to lib/.