Example app for deployment to fission Fission
Kube cluster with fission installed
Python 3, kubectl and fission CLI installed locally
Deploy to fission by creating a spec and applying $ fission spec init
$ fission env create --spec --name python --image fission/python-env
$ fission function create --spec --name hello-you --env python --code app.py --entrypoint main
$ fission route create --spec --method GET --url /hello --function hello-you
Get the IP of the fission router and send a request with timing $ export IP_ADDRESS=$(kubectl --namespace fission get svc router -o=jsonpath='{..ip}')
$ time curl "http://${IP_ADDRESS?}?name=yourname"
0m0.045s
0m0.039s
0m0.053s
Change executor type to newdeploy (which can scale to zero) $ fission function delete hello-you
$ rm -f specs/function*
$ fission function create --spec \
--name hello-you \
--env python \
--code app.py \
--entrypoint main \
--executortype newdeploy \
--minscale 0 \
--maxscale 3 \
--mincpu 100 \
--maxcpu 200 \
--minmemory 128 \
--maxmemory 256 \
--targetcpu 30
$ fission spec apply
Send another request with timing after 10 minutes (or once pods have been terminated) $ time curl "http://${IP_ADDRESS?}?name=yourname"
0m1.065s (a cold start...)
0m0.037s (warm...)
0m0.054s
Look at the HPA using kubectl kubectl -n fission-function get hpa -w
Create another function and schedule it to run every minute $ fission function create --name hello \
--env python \
--code app.py \
--entrypoint main
$ fission tt create --name minute --function hello --cron "@every 1m"