1.
Linux Container
Performance Analysis
Sasha Goldshtein
CTO, Sela Group
goldshtn
goldshtn

2.
The Plan
• This talk explains how to analyze Linux container performance and
which challenges you’ll encounter along the way
• You’ll learn:
To retrieve basic resource utilization per-container
Which deployment strategies to use with container performance tools
To dig into high-CPU issues using perf
To make symbol resolution work across namespaces
To use BPF-based tools with containers
2

3.
Container Resource Utilization
3

4.
Containers Under The Hood in 30 Seconds
CONTROL GROUPS
• Restrict usage (place quotas)
• cpu,cpuacct: used to cap CPU
usage and apply CPU shares
• docker run --cpus --cpu-shares
• memory: used to cap user and
kernel memory usage
• docker run --memory
--kernel-memory
• blkio: used to cap IOPS and
throughput per block device,
and to assign weights (shares)
NAMESPACES
• Restrict visibility
• PID: container gets its own PIDs
• mnt: container gets its own /
• net: container gets its own
network interfaces
• user: container gets its own user
and group ids
• Etc.

5.
USE Checklist for Linux Systems
http://www.brendangregg.com/USEmethod/use-linux.html
CPU
Core Core
LLC
RAM
Memory
controller
GFX
I/O
controller
SSD
E1000
FSB
PCIe
U: perf
U: mpstat -P 0
S: vmstat 1
E: perf
U: sar -n DEV 1
S: ifconfig
E: ifconfig
U: free -m
S: sar -B
E: dmesg
U: iostat 1
S: iostat -xz 1
E: …/ioerr_cntU: perf

6.
Retrieving Container Resource Utilization
• High-level, Docker-specific: docker stats
• htop with cgroup column (highly unwieldy)
• systemd-cgtop
• Execute a command in the container :
• nsenter -t $PID -p -m top
• docker exec -it $CID top
• Control group metrics
• E.g. /sys/fs/cgroup/cpu,cpuacct/*/*/*
• Third-party monitoring solutions: cAdvisor, Intel Snap, PCP, collectd
6

7.
But Beware Of:
• /proc/loadavg showing host statistics
• free showing systemwide memory usage
• iostat showing host block I/O
7

8.
Tool Deployment Strategies
8

9.
Tool Deployment: On The Host
9
User
Kernel
perf_events
node:6 openjdk:8 ubuntu:xenial
# perf record -G …
-g -F 97

10.
Tool Deployment: In Container
10
User
Kernel
perf_events
node:6 openjdk:8
ubuntu:xenial
# perf record -G … -g -F 97
☢฀

11.
Problems
• On the host:
• Need privileged access to the host (most container orchestrators try to
abstract this away from you)
• Tools need to understand container pid namespace, mount namespace, and
other details (discussed later)
• In the container:
• Need to run the container with extra privileges (e.g. for perf: enable
perf_event_open syscall, perf_event_paranoid sysctl)
• Bloats container images with performance tools that are not always used, and
increases attack surface
• Performance tools may be throttled by quotas placed on the container
11

12.
Tool Deployment: Sidecar Container
12
User
Kernel
perf_events
node:6 openjdk:8 ubuntu:xenial
ubuntu:xenial
# perf record -G …
-g -F 97
☢฀

13.
Profiling Containers with perf
13

14.
perf_events Architecture
14
UserKernel
tcp_msgsend
Page fault
LLC miss
sched_switch
perf_events
libc:malloc
mmap buffer
mmap buffer
Real-time
analysis
perf script…perf.data file

15.
Recording CPU Stacks With perf
• To find a CPU bottleneck, record stacks at timed intervals:
# system-wide
perf record -ag -F 97
# specific process
perf record -p 188 -g -F 97
# specific cgroup
perf record -G docker-1ae… -g -F 97
Legend
-a all CPUs
-p specific process
-G specific cgroup
-g capture call stacks
-F frequency of samples (Hz)
-c # of events in each sample

16.
Symbol Resolution Challenges
• Address to module and symbol resolution, dynamic instrumentation
require access to debug information
• Because of mount namespace, container’s binaries and debuginfo are not
visible to host (/lib64/libc.so.6 – what?)
• Need to enter the container’s namespace or share the binaries
16

17.
Perf Maps: Containerized Java/Node/.NET…
• For interpreted or JIT-compiled languages, map files need to be
generated at runtime
$ cat /tmp/perf-1882.map
7f2cd1108880 1e8 Ljava/lang/System;::arraycopy
7f2cd1108c00 200 Ljava/lang/String;::hashCode
7f2cd1109120 2e0 Ljava/lang/String;::indexOf
• When profiling from the host:
• PID namespace – always /tmp/perf-1.map in the container, not on the host
• Mount namespace – container’s /tmp/perf-1.map not visible to host
• perf handles this automatically in Linux 4.13+, as do the BCC tools
(discussed next)
17

18.
Tracing Containers with BPF
18

19.
BPF Tracing
kernel
BPF runtimekprobes
tracepoints
user
control program
BPF program
map
application
USDT
uprobesperf output
installs BPF program and attaches to events
events invoke the BPF program
perf_events
BPF program updates a map or pushes a
new event to a buffer shared with user-space
user-space program is invoked with
data from the shared buffer
user-space program reads statistics
from the map and clears it if necessary
control
program
19

20.
Java/Node/…
Syscall interface
Block I/O Ethernet
Scheduler Mem
Device drivers
Filesystem TCP/IP
CPU
Applications
System libraries
profile
llcstat
hardirqs
softirqs
ttysnoop
runqlat
cpudist
offcputime
offwaketime
cpuunclaimed
memleak
oomkill
slabratetoptcptop
tcplife
tcpconnect
tcpaccept
biotop
biolatency
biosnoop
bitesize
filetop
filelife
fileslower
vfscount
vfsstat
cachestat
cachetop
mountsnoop
*fsslower
*fsdist
dcstat
dcsnoop
mdflush
execsnoop
opensnoop
killsnoop
statsnoop
syncsnoop
setuidsnoop
mysqld_qslower
bashreadline
dbslower
dbstat
mysqlsniff
memleak
sslsniff
gethostlatency
deadlock_detector
ustat
ugc
ucalls
uthreads
uobjnew
uflow
argdist
trace
funccount
funclatency
stackcount
20

21.
Summary
• We have seen:
To retrieve basic resource utilization per-container
Which deployment strategies to use with container performance tools
To dig into high-CPU issues using perf
To make symbol resolution work across namespaces
To use BPF-based tools with containers
21

22.
References
• Container performance analysis
• https://www.slideshare.net/brendangregg/container-performance-analysis
• http://batey.info/docker-jvm-flamegraphs.html
• Performance tooling support
• https://lkml.org/lkml/2017/7/5/771
• https://github.com/iovisor/bcc/pull/1051
• Monitoring tools
• https://github.com/intelsdi-x/snap
• http://pcp.io/docs/guide.html
• https://github.com/bobrik/collectd-docker
• https://github.com/ivotron/docker-perf
22

23.
Thank You!
Slides: https://s.sashag.net/ktlv1217
Demos & labs: https://github.com/goldshtn/linux-tracing-workshop
✁
Sasha Goldshtein
CTO, Sela Group
goldshtn
goldshtn