Skip to content

Latest commit

 

History

History
98 lines (76 loc) · 4.49 KB

scale.md

File metadata and controls

98 lines (76 loc) · 4.49 KB

Scaling and Performance Testing

Whilst The Power is not designed to do scaling/performance/load testing it has some capabilities that can be used for this.

Edit this section in the .gh-api-example.conf file and set the values you want:

## Default values for scripts that generate many of a resource.
# mostly used for testing things on GHES.
number_of_orgs=3
number_of_repos=3
number_of_teams=3
number_of_branches=10
repo_prefix="gestrepo"
org_prefix="fruitbat"
user_prefix="testusertoad"
team_prefix="testteam"
branch_prefix="testbranch"
file_prefix="testfile"

These scripts use connection reuse to speed things up a lot:

python-create-many-issues-connection-reuse.py     
python-create-many-users-connection-reuse.py
python-create-many-orgs-connection-reuse.py       
python-create-many-repos-connection-reuse.py

The connection pooling scripts are currently the fastest way of creating many resources using The Power. You'll need to install urllib3 in your python environment:

python-create-many-repos-connection-reuse-pool.py 
python-create-many-orgs-connection-reuse-pool.py  
python-create-many-users-connection-reuse-pool.py

Pool script taking approx 3 minutes to create 1000 repositories:

python python-create-many-repos-connection-reuse-pool.py
INFO:root:Reading config file: .gh-api-examples.conf
INFO:root:Reading config app_id: 1
INFO:__main__:creating: 1000 repos
INFO:__main__:on host: gm3dmo-0f9aefebf2d0229af.ghe-test.org
INFO:__main__:on org: acme
WARNING:urllib3.connectionpool:Connection pool is full, discarding connection: gm3dmo-0f9aefebf2d0229af.ghe-test.org. Connection pool size: 1
WARNING:urllib3.connectionpool:Connection pool is full, discarding connection: gm3dmo-0f9aefebf2d0229af.ghe-test.org. Connection pool size: 1
WARNING:urllib3.connectionpool:Connection pool is full, discarding connection: gm3dmo-0f9aefebf2d0229af.ghe-test.org. Connection pool size: 1
WARNING:urllib3.connectionpool:Connection pool is full, discarding connection: gm3dmo-0f9aefebf2d0229af.ghe-test.org. Connection pool size: 1
WARNING:urllib3.connectionpool:Connection pool is full, discarding connection: gm3dmo-0f9aefebf2d0229af.ghe-test.org. Connection pool size: 1
WARNING:urllib3.connectionpool:Connection pool is full, discarding connection: gm3dmo-0f9aefebf2d0229af.ghe-test.org. Connection pool size: 1
WARNING:urllib3.connectionpool:Connection pool is full, discarding connection: gm3dmo-0f9aefebf2d0229af.ghe-test.org. Connection pool size: 1
WARNING:urllib3.connectionpool:Connection pool is full, discarding connection: gm3dmo-0f9aefebf2d0229af.ghe-test.org. Connection pool size: 1
WARNING:urllib3.connectionpool:Connection pool is full, discarding connection: gm3dmo-0f9aefebf2d0229af.ghe-test.org. Connection pool size: 1
WARNING:urllib3.connectionpool:Connection pool is full, discarding connection: gm3dmo-0f9aefebf2d0229af.ghe-test.org. Connection pool size: 1
WARNING:urllib3.connectionpool:Connection pool is full, discarding connection: gm3dmo-0f9aefebf2d0229af.ghe-test.org. Connection pool size: 1
elapsed time: 0:03:15.595511 to create 1000 repos

Creating a "large" environment.

  • 30 repos using create-many-repos.sh
./create-many-repos.sh  1.95s user 0.75s system 8% cpu 32.277 total
  • 30 repos using python-create-many-repos-connection-reuse.py
time python3 python-create-many-repos-connection-reuse.py
python3 python-create-many-repos-connection-reuse.py  0.13s user 0.05s system 1% cpu 16.075 total
  • 1000 repos using ./create-many-repos.sh
./create-many-repos.sh  52.60s user 20.04s system 7% cpu 15:24.24 total
  • 1000 repos using python-create-many-repos-connection-reuse.py
time python3 python-create-many-repos-connection-reuse.py --repos 1000 --prefix mtst1 --org acme                       I   12:26:54  
python3 python-create-many-repos-connection-reuse.py --repos 1000 --prefix     1.11s user 0.35s system 0% cpu 9:12.72 total
  • 1000 organizations using python-create-many-orgs-connection-reuse.py
time python3 python-create-many-orgs-connection-reuse.py_id
python3 python-create-many-orgs-connection-reuse.py --orgs 1000 --prefix thou  0.88s user 0.24s system 0% cpu 7:19.29 total

If you keep within the GitHub recommendations about file sizes on your GitHub repos and total size of your repo then you'll probably never need to look at performance. If you don't do that then you probably have a host of other problems.