DOKK / manpages / debian 12 / bpfcc-tools / nodegc-bpfcc.8.en
ugc(8) System Manager's Manual ugc(8)

ugc, javagc, nodegc, pythongc, rubygc - Trace garbage collection events in high-level languages.

javagc [-h] [-v] [-m] [-M MINIMUM] [-F FILTER] pid
nodegc [-h] [-v] [-m] [-M MINIMUM] [-F FILTER] pid
pythongc [-h] [-v] [-m] [-M MINIMUM] [-F FILTER] pid
rubygc [-h] [-v] [-m] [-M MINIMUM] [-F FILTER] pid
ugc [-h] [-v] [-m] [-M MINIMUM] [-F FILTER] [-l {java,node,python,ruby}] pid

This traces garbage collection events as they occur, including their duration and any additional information (such as generation collected or type of GC) provided by the respective language's runtime.

This tool relies on USDT probes embedded in many high-level languages, such as Java, Node, Python, and Ruby. It requires a runtime instrumented with these probes, which in some cases requires building from source with a USDT-specific flag, such as "--enable-dtrace" or "--with-dtrace".

Since this uses BPF, only the root user can use this tool.

CONFIG_BPF and bcc.

Print the resulting BPF program, for debugging purposes.
Print times in milliseconds. The default is microseconds.
Display only collections that are longer than this threshold. The value is given in milliseconds. The default is to display all collections.
Display only collections whose textual description matches (contains) this string. The default is to display all collections. Note that the filtering here is performed in user-space, and not as part of the BPF program. This means that if you have thousands of collection events, specifying this filter will not reduce the amount of data that has to be transferred from the BPF program to the user-space script.
{java,node,python,ruby}
The language to trace.
The process id to trace.

# ugc node 148
milliseconds: # ugc -m java 6004
they are longer than 10ms and have the string "Tenured" in their detailed description: # ugc -M 10 -F Tenured java 6004

The start time of the GC, in seconds from the beginning of the trace.
The duration of the garbage collection event.
The runtime-provided description of this garbage collection event.

Garbage collection events, even if frequent, should not produce a considerable overhead when traced because they are still not very common. Even hundreds of GCs per second (which is a very high rate) will still produce a fairly negligible overhead.

This is from bcc.

https://github.com/iovisor/bcc

Also look in the bcc distribution for a companion _example.txt file containing example usage, output, and commentary for this tool.

Linux

Unstable - in development.

Sasha Goldshtein

trace(8), ustat(8), uobjnew(8)

2018-10-09 USER COMMANDS