Artifact [dac3d7178e]

Not logged in

Artifact dac3d7178e97e4faa2c66f9f34753e1b1afac711223db610b58aeed7f4536fd1:


Spawning LumoSQL Benchmarks on Kubernetes

This service is deployed on the VUB OpenCloudEdge Kubernetes cluster. If you are not associated with VUB then your cluster will look different but the results should be similar.

The generated files are:

If the .sqlite file appears corrupt the job probably hasn't finished running yet. You can test for corruption/incompletion with the command:

sqlite3 $filename.sqlite "PRAGMA integrity_check"

To operate the job system generally:

  1. Set your DNS names in setup.yaml. Probably you want to look for lumosql.opencloudedge.be and change that to something you control, and possibly also change your ingress and PVC settings to something that your cluster supports.
  2. Create swq token: kubectl create secret generic swq-token --from-literal=token=your-very-secret-token
  3. Deploy the thing: kubectl apply -f ./setup.yaml
  4. Send work with curl: curl -L -H "Authorization: Bearer your-very-secret-token" -X PUT lumosql.opencloudedge.be/job/give-your-job-a-sensible-name --data-binary @Makefile.local

To operate the VUB job system specifically:

Linux Image

This system assumes an image. We define ours using the Dockerfile in this directory.

This Dockerfile gives a Linux image satisfying the Build Environment and Dependencies and is installed for every run. The LumoSQL tree is updated with the command "fossil update", the

Makefile.local
supplied via
curl
is copied to the top level and then make is invoked with the parameters in the yaml files. Not-forking does have the ability to upgrade itself, but we do not use that, instead we keep it manually up to date via the Dockerfile image for now.