* ci: parallelize gotest by separating test/cli into own job
split the Go Test workflow into two parallel jobs:
- `unit-tests`: runs unit tests (excluding test/cli)
- `cli-tests`: runs test/cli end-to-end tests
test/cli takes ~3 minutes (~50% of total gotest time), so running
it in parallel should reduce wall-clock CI time by ~1.5-2.5 minutes.
both jobs produce JUnit XML and HTML reports for consistent debugging.
* ci(gotest): reduce noise on test timeout panics
add GOTRACEBACK=single to show only one goroutine stack instead of all
when a test timeout panic occurs. this makes CI output much cleaner
when tests hang.
* fix(ci): prevent stderr from corrupting test JSON output
- remove 2>&1 which mixed "go: downloading" stderr messages into JSON
- add JSON validation before parsing
- print failed test names for easier debugging
* ci(gotest): use gotestsum for human-readable test output
- replace per-package coverage loop with single gotestsum invocation
- both unit-tests and cli-tests now show human-readable output
- simplified coverage collection (single -coverprofile, no gocovmerge)
- clarified step names to indicate they run tests
* ci: fix codecov uploads by adding token
- add CODECOV_TOKEN to gotest.yml and sharness.yml
- update codecov-action to v5.5.2
- add fail_ci_if_error: false for robustness
codecov stopped receiving coverage data ~1 year ago when they
started requiring tokens for public repos
* refactor(make): add test_unit and test_cli targets
- add `make test_unit` for unit tests with coverage (used by CI)
- add `make test_cli` for CLI integration tests (used by CI)
- only disable colors when CI env var is set (local dev gets colors)
- remove legacy targets: test_go_test, test_go_short, test_go_race, test_go_expensive
- update gotest.yml to use make targets instead of inline commands
- add test artifacts to .gitignore
* fix(ci): move client/rpc tests to cli-tests job
client/rpc tests use test/cli/harness which requires the ipfs binary.
Move them from test_unit to test_cli where the binary is built.
also:
- update gotestsum to v1.13.0
- simplify workflow step names
* fix(ci): use build tags when listing test packages
go list needs build tags to properly exclude packages like fuse/mfs
when running with TEST_FUSE=0 (nofuse tag).
* fix(ci): move test/integration to cli-tests job
test/integration tests need the ipfs binary, move them from test_unit
to test_cli.
* fix(test): fix flaky kubo-as-a-library and GetClosestPeers tests
kubo-as-a-library: use `Bootstrap()` instead of raw `Swarm().Connect()`
to fix race condition between swarm connection and bitswap peer
discovery. `Bootstrap()` properly integrates peers into the routing
system, ensuring bitswap learns about connected peers synchronously.
GetClosestPeers: simplify retry logic using `EventuallyWithT` with
10-minute timeout. tests all 4 routing types (`auto`, `autoclient`,
`dht`, `dhtclient`) against real bootstrap peers with patient polling.
* fix(example): use bidirectional Swarm().Connect() for reliable bitswap
- connect nodes bidirectionally (A→B and B→A) to simulate mutual peering
- mutual peering protects connection from resource manager culling
- use port 0 for random available ports (avoids CI conflicts)
- enable LoopbackAddressesOnLanDHT for local testing
- move retry logic to test file using require.Eventually
* fix(ci): add test_examples target and parallel example-tests job
- add `make test_examples` target to mk/golang.mk for consistency with test_unit/test_cli
- move example tests to separate parallel CI job (example-tests)
- example: use Bootstrap() with autoconf.FallbackBootstrapPeers for reliable bitswap
- example: increase context timeout to 10 minutes
- test: add 60s per-request timeout to GetClosestPeers (server has 30s routing timeout)
- test: reduce EventuallyWithT to 3 minutes (locally passes in under 1 minute)
* fix(ci): improve test targets, exclusion patterns, and artifact naming
- define COVERPKG_EXCLUDE and UNIT_EXCLUDE as documented variables
- use grep -vE with single regex instead of multiple grep -v calls
- add mkdir -p before rm to ensure directories exist
- add DEPS_GO dependency to test_cli target
- make CLI test timeout configurable via TEST_CLI_TIMEOUT (default 10m)
- fix test_examples cleanup on failure using subshell
- reduce GetClosestPeers test wait time from 3m to 2m
- rename artifacts to match job names: unit-tests-{junit,html}, cli-tests-{junit,html}
- update cli-tests upload-artifact from v5 to v6
* fix(ci): fix unit test exclusion and speed up example test
- fix UNIT_EXCLUDE regex to match client/rpc at end of path
- remove public bootstrap peers from example (only connect to nodeA)
- example test now runs in ~3s instead of timing out
* fix(test): fix flaky TestAddMultipleGCLive race condition
added time.Sleep after spawning GC goroutines to ensure they reach
GCLock() before the test proceeds. without this, the adder's
maybePauseForGC() might check GCRequested() before GC has even
requested the lock, causing the lock to not be released and GC to
block indefinitely.
this matches the existing pattern in TestAddGCLive which already
had this sleep.
also replaced context.Background() with t.Context() in both
TestAddMultipleGCLive and TestAddGCLive for proper test lifecycle
management.
* fix(example): use test harness settings for reliable CI
the kubo-as-a-library example was flaky on CI. applied test-harness-like
settings that match what transports_test.go uses:
- TCP-only on 127.0.0.1 with random port (no QUIC/UDP)
- explicitly disable non-TCP transports (QUIC, Relay, WebTransport, etc)
- use NilRouterOption (no routing) since we connect peers directly
- bitswap works with directly connected peers without DHT lookups
- 2-minute context timeout
- streaming output in test for debugging
* test: add migration tests for Windows and macOS
- add dedicated CI workflow for migration tests on Windows/macOS
- workflow triggers on migration-related file changes only
* build: remove redundant go version checks
- remove GO_MIN_VERSION and check_go_version scripts
- go.mod already enforces minimum version (go 1.25)
- fixes make build on Windows
* fix: windows migration panic by reading config into memory
fixes migration panic on Windows when upgrading from v0.37 to v0.38
by reading the entire config file into memory before performing atomic
operations. this avoids file locking issues on Windows where open files
cannot be renamed.
also fixes:
- TestRepoDir to set USERPROFILE on Windows (not just HOME)
- CLI migration tests to sanitize directory names (remove colons)
minimal fix that solves the "panic: error can't be dealt with
transactionally: Access is denied" error without adding unnecessary
platform-specific complexity.
* fix: set PATH for CLI migration tests in CI
the CLI tests need the built ipfs binary to be in PATH
* fix: use ipfs shutdown for graceful daemon termination in tests
replaces platform-specific signal handling with ipfs shutdown command
which works consistently across all platforms including Windows
* fix: isolate PATH modifications in parallel migration tests
tests running in parallel with t.Parallel() were interfering with each
other through global PATH modifications via os.Setenv(). this caused
tests to download real migration binaries instead of using mocks,
leading to Windows failures due to path separator issues in external tools.
now each test builds its own custom PATH and passes it explicitly to
commands, preventing interference between parallel tests.
* chore: improve error messages in WithBackup
* fix: Windows CI migration test failures
- add .exe extension to mock migration binaries on Windows
- handle repo lock file properly in mock migration binary
- ensure lock is created and removed to prevent conflicts
* refactor: align atomicfile error handling with fs-repo-migrations
- check close error in Abort() before attempting removal
- leave temp file on rename failure for debugging (like fs-repo-15-to-16)
- improves consistency with external migration implementations
* fix: use req.Context in repo migrate to avoid double-lock
The repo migrate command was calling cctx.Context() which has a hidden
side effect: it lazily constructs the IPFS node by calling GetNode(),
which opens the repository and acquires repo.lock. When migrations then
tried to acquire the same lock, it failed with "lock is already held by us"
because go4.org/lock tracks locks per-process in a global map.
The fix uses req.Context instead, which is a plain context.Context with
no side effects. This provides what migrations need (cancellation handling)
without triggering node construction or repo opening.
Context types explained:
- req.Context: Standard Go context for request lifetime, cancellation,
and timeouts. No side effects.
- cctx.Context(): Kubo-specific method that lazily constructs the full
IPFS node (opens repo, acquires lock, initializes subsystems). Returns
the node's internal context.
Why req.Context is correct here:
- Migrations work on raw filesystem (only need ConfigRoot path)
- Command has SetDoesNotUseRepo(true) - doesn't need running node
- Migrations handle their own locking via lockfile.Lock()
- Need cancellation support but not node lifecycle
The bug only appeared with embedded migrations (v16+) because they run
in-process. External migrations (pre-v16) were separate processes, so
each had isolated state. Sequential migrations (forward then backward)
in the same process exposed this latent double-lock issue.
Also adds repo.lock acquisition to RunEmbeddedMigrations to prevent
concurrent migration access, and removes the now-unnecessary daemon
lock check from the migrate command handler.
* fix: use req.Context for migrations and autoconf in daemon startup
daemon.go was incorrectly using cctx.Context() in two critical places:
1. Line 337: migrations call - cctx.Context() triggers GetNode() which
opens the repo and acquires repo.lock BEFORE migrations run, causing
"lock is already held by us" errors when migrations try to lock
2. Line 390: autoconf client.Start() - uses context for HTTP timeouts
and background updater lifecycle, doesn't need node construction
Both now use req.Context (plain Go context) which provides:
- request lifetime and cancellation
- no side effects (doesn't construct node or open repo)
- correct lifecycle for HTTP requests and background goroutines
* ci: optimize build workflows
- use go version from go.mod instead of hardcoding
- group platforms by OS for parallel builds
- remove legacy try-build targets
* fix: checkout before setup-go in all workflows
setup-go needs go.mod to be present, so checkout must happen first
* chore: remove deprecated // +build syntax
go 1.17+ uses //go:build, the old syntax is no longer needed
* simplify: remove nofuse tag from CI workflows
- workflows now rely on platform build constraints
- keep make nofuse target for manual builds
- remove unused appveyor.yml
* ci: remove legacy travis variable and fix gateway-conformance
- remove TRAVIS env variable from 4 workflows
- fix gateway-conformance checkout path to match working-directory
- replace deprecated cache-go-action with built-in setup-go caching
* fix(ci): pass -buildvcs=false to go list in Makefile
Docker builds were failing because go list commands in mk/golang.mk
didn't inherit GOFLAGS from the environment, causing empty package
names when .git directory was incomplete.
This issue always existed but was masked until recent changes in
GitHub Actions runners or Docker golang image exposed it.
* fix(ci): use modern ENV key=value format in Dockerfile
Fixes Docker build deprecation warnings.
* add deprecation warning when tracer plugins are loaded
* add response format attribute to span in gateway handler
* add note about tracing's experimental status in godoc
* add nil check for TTL when adding name span attrs
* add basic sharness test for integration with otel collector
* add nil check in UnixFSAPI.processLink
* test: sharness check all json objs for swarm span
* add env var docs to docs/environment-variables.md
* chore: pin the otel collector version
* add tracing spans per response type (#8841)
* docs: tracing with jaeger-ui
Co-authored-by: Marcin Rataj <lidel@lidel.org>
Set supported platforms correctly by go version (no darwin-386 for >=1.15, no darwin-arm64 unless
>=1.16), make check-go-version work on mac
License: MIT
Signed-off-by: hannahhoward <hannah@hannahhoward.net>
That way we can combine our GOFLAGS with the user specified ones.
This also switches to the new -trimpath build flag for reproducible builds.
fixes#7475
Go 1.14 has a timer reset deadlock (https://github.com/golang/go/issues/38070).
This also downgrades quic-go until either a go patch release fixes this issue or
a version of quic-go is released that works with go 1.13.
A dependency of ours now requires 1.14, let's be more explicit ourselves
../../go/pkg/mod/github.com/lucas-clemente/quic-go@v0.15.2/interface.go:54:2: duplicate method StreamID
../../go/pkg/mod/github.com/lucas-clemente/quic-go@v0.15.2/stream.go:44:2: duplicate method StreamID
note: module requires Go 1.14
The usage of a native 'go' command has been replaced with a make &
environment variable $GOCC. This enables building with multiple go
versions on a single machine as documented:
* https://golang.org/doc/install#extra_versions
This enables the usage of:
```bash
$ make install
$ # OR
$ GOCC=go1.12.3 make install
$ # OR
$ GOCC=go1.12.4 make install
```
And the build and test tools now pick up on this change
On branch go-version-check
Changes to be committed:
modified: Rules.mk
modified: bin/check_go_version
modified: bin/dist_get
modified: bin/maketarball.sh
modified: coverage/Rules.mk
modified: mk/golang.mk
modified: mk/tarball.mk
License: MIT
Signed-off-by: Chris Buesser <christopher.buesser@gmail.com>
With this command we will get hash information as well as information if
build was made using uncommited files.
License: MIT
Signed-off-by: Jakub Sztandera <kubuxu@protonmail.ch>
