mirror of
https://github.com/hanzoai/benchmarks.git
synced 2026-10-30 14:59:11 +00:00
zeekayandHanzo Dev
377c4f4416
Twelve lanes in the order that needs the least, so the first failure names the smallest missing piece: node, then go, then the cloud source, then fetched data. A lane whose data nobody has fetched is SKIPPED and says which command fetches it, counted separately from the passes. Twelve ok over four silent skips is the same green a broken suite prints, and this suite has spent a week finding exactly that shape. It names the tree the binary lanes build from, above their results. Two different repositories check out as `cloud`; against the wrong one all five run and fail, which reads as five broken lanes rather than one wrong path. With the line printed, `hanzo-inc/cloud · c17b10d8` sits directly above four failures and the diagnosis is immediate. Verified from a fresh clone: 12 ok, 0 failed, 2 skipped with the source named. Co-authored-by: Hanzo Dev <[email protected]>
29 lines
905 B
Makefile
29 lines
905 B
Makefile
GO ?= go
|
|
GOFLAGS ?= -mod=mod
|
|
export GOPRIVATE = github.com/hanzoai
|
|
|
|
.PHONY: all check test bench scale k3s-up k3s-down help
|
|
|
|
help: ## list targets
|
|
@grep -hE '^[a-z-]+:.*##' $(MAKEFILE_LIST) | sed -E 's/:.*## / — /' | sort
|
|
|
|
check: ## does this checkout work — every lane, skipping what it cannot reach
|
|
@bash check.sh
|
|
|
|
all: test bench ## property tests + throughput
|
|
|
|
scale: ## horizontal-scale proof: 1M tenants across N=3/10/100 pods (prints)
|
|
$(GO) test $(GOFLAGS) ./cloud/shard/ -run 'ScaleProof|Deterministic|Reshuffle' -v -count=1
|
|
|
|
bench: ## routing throughput (ns/op, allocs) across ring sizes
|
|
$(GO) test $(GOFLAGS) ./cloud/... -bench=. -benchmem -run '^$$'
|
|
|
|
test: ## all cluster-free property tests
|
|
$(GO) test $(GOFLAGS) ./... -count=1
|
|
|
|
k3s-up: ## integration: stand up the cloud StatefulSet N=3 in local k3s
|
|
./k3s/up.sh
|
|
|
|
k3s-down: ## tear down the local k3s shard stack
|
|
./k3s/down.sh
|