-
-
Notifications
You must be signed in to change notification settings - Fork 191
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
GitlabCI fails constantly once cache is made. Any help welcome #768
Comments
Additional steps to make the local caches specialized would be desired, just like attempted for CircleCI at #769 |
@szszszsz : were you able to go around that problem on your side? Do you know anybody that has more experience on that level? How much would it cost to fix this? |
Really not understand why CircleCI cache is 5.1GiB:
And why GitlabCI is 14GiB (for less boards..., same paths...):
Of course, deflating 14Gb on top of a docker image might break docker reserved space. The question is why such big discrepency between the two CI environements, where CircleCI cache ressembles the size of local builds, where GitlabCI cache size..... doesn't make any sense, like if files were archived multiple times. Symlinks not followed and archived multiple times? Other hypothesis? |
Note: GitlabCI now implements a caching system based on file hashes. So CircleCI equivalent can be made. I reply to
The number of files matches what is observed locally. The problem here seems to be that the cache itself is not really compressed. It was not a problem before, but it seems that GitLabCI changed their internals, where cache.zip doesn't seem to imply strong compression/CircleCI implements it, which explains the difference of size of the cache, but not why it is incapable of decompressing without causing disk space issue. Reasoning: If the cahce built from local build created files were able to be expending by compilation, there should not be a problem expending that cache locally on top of a docker image. So probably a problem of paths and misunderstanding on my side of the paths implied. Will do WIP based on that with trial and errors |
@szszszsz @jan23 @MrChromebox, @daringer @alex-nitrokey : any help welcome, as always. |
Update: Is where the cache is extracted. Will try to figure out how the tasks can be reordered in BitlabCI conf, since:
|
…. WiP attacking linuxboot#768 Starting with only building qemu-coreboot board to speed tests up.
Since its two steps, I have now problems downloading cache of musl-cross when there is no matches for modules+patches. |
Swtiched to CircleCI years ago |
You can see CI errors when looking at commit logs on master.
One example of such failing build.
Output:
The text was updated successfully, but these errors were encountered: