-
Notifications
You must be signed in to change notification settings - Fork 18
/
Copy pathNOTES-ON-PROCEDURE-CONFIG.txt
47 lines (28 loc) · 2.09 KB
/
NOTES-ON-PROCEDURE-CONFIG.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
NEW BENCHMARK CONFIG:
# Run 1 & 2, aws to aws: AWS server to server, 2x AWS c4.large instances, running up to date Amazon linux
Installation from ../aws-setup.sh
Benchmark results (CSV) saved to aws-to-aws-nginx-http.csv
Run log (terminal output) saved to aws-to-aws-nginx-http-log.txt
Benchmark results (CSV) saved to aws-to-aws-nginx-https.csv
Run log (terminal output) saved to aws-to-aws-nginx-https-log.txt
Server vmstat (every 10 sec with timestamp) - saved to vmstat-log-server-hit-by-client-https-test.txt
-- DONE WITH COMMIT HASH 487d68134db541c01eaa977ac43aea64ba5baaab --
Image v2.0 (pulled from dockerhub)
# Client commands
python benchmark.py --cycles 10000 --url http://172.31.22.99:8080 --benchmark-type size --output-file new-aws-results/aws-to-aws-nginx-http.csv | tee new-aws-results/aws-to-aws-nginx-http-log.txt
(and same, but https & port 4443)
Top reveals that during HTTPS tests, CPU runs ~50%+ for python process
# Server:
Start server:
sudo docker run --rm -i -t -w /tmp -p 4443:443 -p 8080:80 svanoort/client-flask-demo:2.0
vmstat 1 reveals that CPU is mostly in wait on I/O mode on the server, sometimes CPU goes as high as 40%
Saved results by: vmstat -t 10 | tee https-vmstat.txt
# TESTING CLIENT-LOOPBACK ON ITSELF
DONE WITH COMMIT HASH 7af6ad13ea2f32c84b6dd55456dff2eda35801af
(this includes fix for urllib3 issues, and ended up commenting out urllib3 & urllib bits because crashy with verify errors)
In one terminal, start client:
sudo docker run --rm -i -t -w /tmp -p 4443:443 -p 8080:80 svanoort/client-flask-demo:2.0
In another terminal, benchmark it fully without HTTPS, using 1024B requests
python benchmark.py --cycles 10000 --url https://127.0.0.1:8080/length/1024 --output-file new-aws-results/aws-loopback-full-nginx-http.csv | tee new-aws-results/aws-loopback-full-nginx-http-log.txt
python benchmark.py --cycles 10000 --url https://127.0.0.1:4443/length/1024 --output-file new-aws-results/aws-loopback-full-nginx-https.csv | tee new-aws-results/aws-loopback-full-nginx-https-log.txt
NOTE: in top, load average sometimes hits 2.0, python benchmark consistently hits 60%