1ucalls(8) System Manager's Manual ucalls(8)
2
3
4
6 ucalls, javacalls, perlcalls, phpcalls, pythoncalls, rubycalls,
7 tclcalls - Summarize method calls from high-level languages and Linux
8 syscalls.
9
11 javacalls [-h] [-T TOP] [-L] [-S] [-v] [-m] pid [interval]
12 perlcalls [-h] [-T TOP] [-L] [-S] [-v] [-m] pid [interval]
13 phpcalls [-h] [-T TOP] [-L] [-S] [-v] [-m] pid [interval]
14 pythoncalls [-h] [-T TOP] [-L] [-S] [-v] [-m] pid [interval]
15 rubycalls [-h] [-T TOP] [-L] [-S] [-v] [-m] pid [interval]
16 tclcalls [-h] [-T TOP] [-L] [-S] [-v] [-m] pid [interval]
17 ucalls [-l {java,perl,php,python,ruby}] [-h] [-T TOP] [-L] [-S] [-v]
18 [-m] pid [interval]
19
21 This tool summarizes method calls from high-level languages such as
22 Java, Perl, PHP, Python, Ruby, and Tcl. It can also trace Linux system
23 calls. Whenever a method is invoked, ucalls records the call count and
24 optionally the method's execution time (latency) and displays a sum‐
25 mary.
26
27 This uses in-kernel eBPF maps to store per process summaries for effi‐
28 ciency.
29
30 This tool relies on USDT probes embedded in many high-level languages,
31 such as Java, Perl, PHP, Python, Ruby, and Tcl. It requires a runtime
32 instrumented with these probes, which in some cases requires building
33 from source with a USDT-specific flag, such as "--enable-dtrace" or
34 "--with-dtrace". For Java, method probes are not enabled by default,
35 and can be turned on by running the Java process with the
36 "-XX:+ExtendedDTraceProbes" flag. For PHP processes, the environment
37 variable USE_ZEND_DTRACE must be set to 1.
38
39 Since this uses BPF, only the root user can use this tool.
40
42 CONFIG_BPF and bcc.
43
45 -l {java,perl,php,python,ruby,tcl}
46 The language to trace. If not provided, only syscalls are traced
47 (when the -S option is used).
48
49 -T TOP Print only the top methods by frequency or latency.
50
51 -L Collect method invocation latency (duration).
52
53 -S Collect Linux syscalls frequency and timing.
54
55 -v Print the resulting BPF program, for debugging purposes.
56
57 -m Print times in milliseconds (the default is microseconds).
58
59 pid The process id to trace.
60
61 interval
62 Print summary after this number of seconds and then exit. By
63 default, wait for Ctrl+C to terminate.
64
66 Trace the top 10 Ruby method calls:
67 # ucalls -T 10 -l ruby 1344
68
69 Trace Python method calls and Linux syscalls including latency in mil‐
70 liseconds:
71 # ucalls -l python -mL 2020
72
73 Trace only syscalls and print a summary after 10 seconds:
74 # ucalls -S 788 10
75
77 Tracing individual method calls will produce a considerable overhead in
78 all high-level languages. For languages with just-in-time compilation,
79 such as Java, the overhead can be more considerable than for inter‐
80 preted languages. On the other hand, syscall tracing will typically be
81 tolerable for most processes, unless they have a very unusual rate of
82 system calls.
83
85 This is from bcc.
86
87 https://github.com/iovisor/bcc
88
89 Also look in the bcc distribution for a companion _example.txt file
90 containing example usage, output, and commentary for this tool.
91
93 Linux
94
96 Unstable - in development.
97
99 Sasha Goldshtein
100
102 ustat(8), argdist(8)
103
104
105
106USER COMMANDS 2018-10-09 ucalls(8)