https://linen.dev logo
Join Slack
Powered by
is anyone already looking into the <build issues>?...
# contributing-to-airbyte
c
is anyone already looking into the build issues?
Copy code
AcceptanceTests > initializationError FAILED
    java.lang.IllegalStateException: Check failed: Docker environment should have more than 2GB free disk space
        at org.testcontainers.DockerClientFactory.check(DockerClientFactory.java:298)
        at org.testcontainers.DockerClientFactory.checkDiskSpace(DockerClientFactory.java:287)
        at org.testcontainers.DockerClientFactory.client(DockerClientFactory.java:224)
        at org.testcontainers.LazyDockerClient.getDockerClient(LazyDockerClient.java:14)
        at org.testcontainers.LazyDockerClient.authConfig(LazyDockerClient.java:12)
        at org.testcontainers.containers.GenericContainer.start(GenericContainer.java:310)
        at io.airbyte.test.acceptance.AcceptanceTests.init(AcceptanceTests.java:143)
u
not looking into it right now but my guess is that the images we install during CI has grown too large for Github’s liking
u
u
it is very possible we’re building > 14gb worth of images (I guess 12, since docker needs 2gb free)
u
one fix could be to remove all standard source test images for connectors that don’t use them
u
(AFAIK only google sheets needs the non-file-based standard tests)
u
u
this commit seems to have pushed it over the edge?
u
i'm actually not sure if that makes sense or just happens to be when the disk died.
u
i think it’s the latter
u
that commit doesn’t touch the core codebase, the main CI build, or anything that gets put into a docker image
u
it does though, no?
u
Copy code
uses: ridedott/merge-me-action@v1
u
Copy code
uses: rtCamp/action-slack-notify@master
u
these are newly added docker images that run in the CI, no?
u
but those only get run when you run the 
publish
command not when you run the main build
u
so like, they shouldn’t be getting installed into the 
gradle.yml
build environment?
u
ah. what you say makes sense.
u
also the only net package we’re adding here is 
merge-me-action
. the other is already used in 
gradle.yml
u
dunno. maybe something doesn't work the way we think it does?
u
after that commit every subsequent commit on master fails.
u
publish workflows from braches are all succeding.
u
maybe the cached contents are too large on master?
u
so either the master build environments aren't hermetic that machine is filled up or something. or that stuff is geting pulled into the master build env somehow?
u
is there an easy way to test what you just mentione jared?
u
we could add a step that outputs the size of cached directories
u
after it loads them
u
or we could nuke by changing 
CACHE_VERSION
and check
u
i defer to you. you've played with this much more than i have.
u
i’ll just nuke for now
u
rerunning the master build
u
thanks.
u
still failing for the same reason
u
not sure why
u
is the caching separate by workflow? e.g. main build caching separate from that of PR builds?
u
yeah
u
u
Thanks.
u
what's the next thing to try? just remove all of the caching and see what happens? i can do this if you feel like that's what we should try next.
u
Next thing to try is to find out what is taking up all the space in a step in the gradle workflow and output it
u
We need to find what is taking up all the space and why that changed recently
u
I’m not planning on looking into this today. Need to focus on Temporal stuff.
u
I also get this error once in a while locally.., i usualky run 
docker system prune
to get out of it but it's annoying...
u
I am going to run this experiment to see if it helps: https://github.com/airbytehq/airbyte/pull/1998
u
i am choosing this experiment because i could do it in <1 minute of work. The logging filesystem output will take me time to figure out. assuming this does not fix anything (which i think is the expectation) then we will need to dedicate some real time to this.
Open in Slack
PreviousNext
Powered by