S3 binary cache no respecting PROXY env variables #433
Labels
No labels
Area/build-packaging
Area/cli
Area/evaluator
Area/fetching
Area/flakes
Area/language
Area/profiles
Area/protocol
Area/releng
Area/remote-builds
Area/repl
Area/store
bug
crash 💥
Cross Compilation
devx
docs
Downstream Dependents
E/easy
E/hard
E/help wanted
E/reproducible
E/requires rearchitecture
imported
Needs Langver
OS/Linux
OS/macOS
performance
regression
release-blocker
RFD
stability
Status
blocked
Status
invalid
Status
postponed
Status
wontfix
testing
testing/flakey
ux
No milestone
No project
No assignees
3 participants
Notifications
Due date
No due date set.
Dependencies
No dependencies set.
Reference: lix-project/lix#433
Loading…
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Describe the bug
In the original nix we have the same bug
when running nix build airgapped with a PROXY used to call internet all aws-sdk-cpp requests are done PROXY-less, so the host tries to join internet directly, which fails
Steps To Reproduce
block internet traffic except to the proxy host
allow anything on the proxy
try to build / fetch some cached biary pkgs
Expected behavior
uses cache corectly
nix --version
outputtried with Nix officiel and Lix beta / rc version
Additional context
not sure if this is a too much breaking change or if it can be implemented easily, so I made this issue to have more context / feedback on this behavior
AWS SDK added a ClientConfig option called allowSystemProxy, which let curl use PROXYs env variables
can be "fixed" by setting the option to True somewhere near there :
ref<Aws::Client::ClientConfiguration> S3Helper::makeConfig(
const std::string & region,
const std::string & scheme,
const std::string & endpoint)
{
initAWS();
auto res = make_ref<Aws::Client::ClientConfiguration>();
res->region = region;
if (!scheme.empty()) {
res->scheme = Aws::Http::SchemeMapper::FromString(scheme.c_str());
}
if (!endpoint.empty()) {
res->endpointOverride = endpoint;
}
res->requestTimeoutMs = 600 * 1000;
res->connectTimeoutMs = 5 * 1000;
res->retryStrategy = std::make_shared<RetryStrategy>();
res->caFile = settings.caFile;
return res;
}
Didn't notice this report. As far as I know the fact that we are not setting allowSystemProxy is a bug. I'll go submit a change to just set that; I cannot imagine a good reason that the proxy should not be used for s3.
This issue was mentioned on Gerrit on the following CLs: