[feature] Inherit resource limits from cgroups (#1336)

When GTS is running in a container runtime which has configured CPU or
memory limits or under an init system that uses cgroups to impose CPU
and memory limits the values the Go runtime sees for GOMAXPROCS and
GOMEMLIMIT are still based on the host resources, not the cgroup.

At least for the throttling middlewares which use GOMAXPROCS to
configure their queue size, this can result in GTS running with values
too big compared to the resources that will actuall be available to it.

This introduces 2 dependencies which can pick up resource contraints
from the current cgroup and tune the Go runtime accordingly. This should
result in the different queues being appropriately sized and in general
more predictable performance. These dependencies are a no-op on
non-Linux systems or if running in a cgroup that doesn't set a limit on
CPU or memory.

The automatic tuning of GOMEMLIMIT can be disabled by either explicitly
setting GOMEMLIMIT yourself or by setting AUTOMEMLIMIT=off. The
automatic tuning of GOMAXPROCS can similarly be counteracted by setting
GOMAXPROCS yourself.

vendor/github.com/cilium/ebpf/link/iter.go

@@ -0,0 +1,91 @@
+package link
+
+import (
+	"fmt"
+	"io"
+
+	"github.com/cilium/ebpf"
+)
+
+type IterOptions struct {
+	// Program must be of type Tracing with attach type
+	// AttachTraceIter. The kind of iterator to attach to is
+	// determined at load time via the AttachTo field.
+	//
+	// AttachTo requires the kernel to include BTF of itself,
+	// and it to be compiled with a recent pahole (>= 1.16).
+	Program *ebpf.Program
+}
+
+// AttachIter attaches a BPF seq_file iterator.
+func AttachIter(opts IterOptions) (*Iter, error) {
+	link, err := AttachRawLink(RawLinkOptions{
+		Program: opts.Program,
+		Attach:  ebpf.AttachTraceIter,
+	})
+	if err != nil {
+		return nil, fmt.Errorf("can't link iterator: %w", err)
+	}
+
+	return &Iter{link}, err
+}
+
+// LoadPinnedIter loads a pinned iterator from a bpffs.
+func LoadPinnedIter(fileName string) (*Iter, error) {
+	link, err := LoadPinnedRawLink(fileName)
+	if err != nil {
+		return nil, err
+	}
+
+	return &Iter{link}, err
+}
+
+// Iter represents an attached bpf_iter.
+type Iter struct {
+	link *RawLink
+}
+
+var _ Link = (*Iter)(nil)
+
+func (it *Iter) isLink() {}
+
+// FD returns the underlying file descriptor.
+func (it *Iter) FD() int {
+	return it.link.FD()
+}
+
+// Close implements Link.
+func (it *Iter) Close() error {
+	return it.link.Close()
+}
+
+// Pin implements Link.
+func (it *Iter) Pin(fileName string) error {
+	return it.link.Pin(fileName)
+}
+
+// Update implements Link.
+func (it *Iter) Update(new *ebpf.Program) error {
+	return it.link.Update(new)
+}
+
+// Open creates a new instance of the iterator.
+//
+// Reading from the returned reader triggers the BPF program.
+func (it *Iter) Open() (io.ReadCloser, error) {
+	linkFd, err := it.link.fd.Value()
+	if err != nil {
+		return nil, err
+	}
+
+	attr := &bpfIterCreateAttr{
+		linkFd: linkFd,
+	}
+
+	fd, err := bpfIterCreate(attr)
+	if err != nil {
+		return nil, fmt.Errorf("can't create iterator: %w", err)
+	}
+
+	return fd.File("bpf_iter"), nil
+}