bcc-drsnoop - Man Page

drsnoop.py [-h] [-T] [-U] [-p PID] [-t TID] [-u UID] [-d DURATION] [-n name] [-v]

drsnoop trace direct reclaim events, showing which processes are allocing pages  with direct reclaiming. This can be useful for discovering when allocstall (/p- roc/vmstat) continues to increase, whether it is caused by some critical proc- esses or not.

This works by tracing the direct reclaim events using kernel tracepoints.

This makes use of a Linux 4.4 feature (bpf_perf_event_output()); for kernels older than 4.4, see the version under tools/old, which uses an older mechanism.

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

CONFIG_BPF and bcc.

-h

Print usage message.

-T

Include a timestamp column.

-U

Show UID.

-p PID

Trace this process ID only (filtered in-kernel).

-t TID

Trace this thread ID only (filtered in-kernel).

-u UID

Trace this UID only (filtered in-kernel).

-d DURATION

Total duration of trace in seconds.

-n name

Only print processes where its name partially matches 'name' -v verbose          Run in verbose mode. Will output system memory state

Trace all direct reclaim events:

# drsnoop

Trace all direct reclaim events, for 10 seconds only:

# drsnoop -d 10

Trace all direct reclaim events, and include timestamps:

# drsnoop -T

Show UID:

# drsnoop -U

Trace PID 181 only:

# drsnoop -p 181

Trace UID 1000 only:

# drsnoop -u 1000

Trace all direct reclaim events from processes where its name partially match-

es 'mond': # drnsnoop -n mond

TIME(s)

Time of the call, in seconds.

UID

User ID

PID

Process ID

TID

Thread ID

COMM

Process name

This traces the kernel direct reclaim tracepoints and prints output for each  event. As the rate of this is generally expected to be low (< 1000/s), the  overhead is also expected to be negligible.

This is from bcc.

https://github.com/iovisor/bcc

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

Linux

Unstable - in development.

Wenbo Zhang