In keep-ecdsa we've seen a number of workflows failing with No space lefton device error.
The problem was a combination of a big cache size and significant memory
allocation after building Docker image.
According to moby/buildkit#1850:
"At the moment caches are copied over the existing cache so it keeps
growing".
As a temporary fix (until issue gets fixed by GH), we introduced a step
that moves the cache (similarily to how it's described in
moby/buildkit#1896) instead of copying.
Although we havent seen the problems with cache size yet in other
projects than keep-ecdsa, we're applying the solution across the
repositories to decrease the likelyhood of encountering problem in the
future.
In
keep-ecdsa
we've seen a number of workflows failing withNo space lefton device
error. The problem was a combination of a big cache size and significant memory allocation after building Docker image. According to moby/buildkit#1850: "At the moment caches are copied over the existing cache so it keeps growing". As a temporary fix (until issue gets fixed by GH), we introduced a step that moves the cache (similarily to how it's described in moby/buildkit#1896) instead of copying. Although we havent seen the problems with cache size yet in other projects thankeep-ecdsa
, we're applying the solution across the repositories to decrease the likelyhood of encountering problem in the future.See also: https://github.com/keep-network/keep-ecdsa/pull/822 https://github.com/keep-network/keep-core/pull/2501 https://github.com/keep-network/tbtc/pull/807