Our project unfortunately requires ~45–60 mins worth, i.e. 1–1.5 GB of cacheable output per run, which makes caching very important for us. We have separate runs (with incompatible caches) for { CI, releases } x { Linux, macOS, Windows } x {}, so as you can imagine, we are hitting the limit and invalidating a lot, leading to way more CPU time used during runs due to cache misses.
For a while we were at 40GB of 10GB on the cache quota, and now we are at 9.1GB of 10GB, I guess that the cache eviction was made a little stricter recently.
Are there any workarounds, e.g. borrow space from other repos which hardly use any cache? Or a paid option?
I have seen the runs-on.com suggestion, but it seems challenging to set up, and I also couldn't see that it supported more than Linux, though I'm not sure.
Our project unfortunately requires ~45–60 mins worth, i.e. 1–1.5 GB of cacheable output per run, which makes caching very important for us. We have separate runs (with incompatible caches) for { CI, releases } x { Linux, macOS, Windows } x {}, so as you can imagine, we are hitting the limit and invalidating a lot, leading to way more CPU time used during runs due to cache misses.
For a while we were at 40GB of 10GB on the cache quota, and now we are at 9.1GB of 10GB, I guess that the cache eviction was made a little stricter recently.
Are there any workarounds, e.g. borrow space from other repos which hardly use any cache? Or a paid option?
I have seen the runs-on.com suggestion, but it seems challenging to set up, and I also couldn't see that it supported more than Linux, though I'm not sure.
Thanks in advance!!