After I turned off all caching, all build contexts were still cached in the /kaniko directory

[hugtale]

All relevant versions:
kaniko: v1.22.0-debug
tekton: v0.57.0
kubernetes: v1.28.3-aliyun.1
containerd: 1.6.28
host os: aliyun_3_9_x64_20G_alibase_20231219.vhd

command options :
--cache=false
--cache-run-layers=false
--cache-copy-layers=false
--compressed-caching=false
--cleanup
--context=$(workspaces.source.path)
--dockerfile=$(workspaces.source.path)/Dockerfile
--destination=${image}
--digest-file=$(results.IMAGE_DIGEST.path)

others:
workingDir: /
runAs: root
workspaces.source: /workerspace/source
results.IMAGE_DIGEST.path: /tekton/results/IMAGE_DIGEST

The above is the build environment and command line options I used.

I'm using kaniko in tekton and I have no need to use kaniko cache.

But I still found that there are built contexts cache in the /kaniko directory. My common technical architecture is Java, and each jar package is more than 100M. If there is such a context cache problem, once the number of tekton tasks reaches hundreds or thousands or more, my k8s host disk will be full. At present, my temporary method is to clean up tekton tasks regularly, but I still want to keep most of the task containers because I want to see the historical container logs.

So, please help to see if what I reported is a problem. If so, please help fix it as soon as possible. Thank you very much!

[hugtale]

This picture is seen on the host machine by searching the upperDir directory of the container. At first, I saw the abnormal disk usage problem of the stopped tekton task container through "crictl ps -a".

[hugtale]

Because the "--cleanup" option was used earlier, I can only clean this context cache in this way at present. Hope it gets fixed soon, thank you very much!