bcc-statsnoop(8)

1statsnoop(8)                System Manager's Manual               statsnoop(8)
2
3
4

NAME

6       statsnoop - Trace stat() syscalls. Uses Linux eBPF/bcc.
7

SYNOPSIS

9       statsnoop [-h] [-t] [-x] [-p PID]
10

DESCRIPTION

12       statsnoop traces the different stat() syscalls, showing which processes
13       are attempting to read information about which files. This can be  use‐
14       ful  for determining the location of config and log files, or for trou‐
15       bleshooting applications that are failing, especially on startup.
16
17       This works by tracing various kernel sys_stat() functions using dynamic
18       tracing,  and  will  need  updating to match any changes to these func‐
19       tions.
20
21       This makes use of a Linux 4.5  feature  (bpf_perf_event_output());  for
22       kernels  older than 4.5, see the version under tools/old, which uses an
23       older mechanism.
24
25       Since this uses BPF, only the root user can use this tool.
26

REQUIREMENTS

28       CONFIG_BPF and bcc.
29

OPTIONS

31       -h     Print usage message.
32
33       -t     Include a timestamp column: in seconds since  the  first  event,
34              with decimal places.
35
36       -x     Only print failed stats.
37
38       -p PID Trace this process ID only (filtered in-kernel).
39

EXAMPLES

41       Trace all stat() syscalls:
42              # statsnoop
43
44       Trace all stat() syscalls, and include timestamps:
45              # statsnoop -t
46
47       Trace only stat() syscalls that failed:
48              # statsnoop -x
49
50       Trace PID 181 only:
51              # statsnoop -p 181
52

FIELDS

54       TIME(s)
55              Time of the call, in seconds.
56
57       PID    Process ID
58
59       COMM   Process name
60
61       FD     File descriptor (if success), or -1 (if failed)
62
63       ERR    Error number (see the system's errno.h)
64
65       PATH   Open path
66

OVERHEAD

68       This  traces the kernel stat function and prints output for each event.
69       As the rate of this is generally expected to be  low  (<  1000/s),  the
70       overhead  is also expected to be negligible. If you have an application
71       that is calling a high rate of stat()s, then test and understand  over‐
72       head before use.
73

SOURCE

75       This is from bcc.
76
77              https://github.com/iovisor/bcc
78
79       Also  look  in  the bcc distribution for a companion _examples.txt file
80       containing example usage, output, and commentary for this tool.
81

OS

83       Linux
84

STABILITY

86       Unstable - in development.
87

AUTHOR

89       Brendan Gregg
90