This adds a plugin interface that lets the plugin modify the fx
options that are passed to fx when the app is initialized. This means
plugins can inject their own implementations of IPFS interfaces. This
enables granular customization of go-ipfs behavior by plugins, such
as:
- Bitswap with custom filters (e.g. for CID blocking) Custom interface
- implementations such as Pinner or DAGService
- Dynamic configuration of libp2p ...
One downside of this is that we're exposing the entire dependency
graph, init hooks, initialization, etc. to users, so this comes with a
caveat that we reserve the right to make breaking changes to the graph
structure and initialization logic (although this historically happens
rarely). If these things are changed, we should mention them in
release notes and changelogs though, since they could impact users of
this plugin interface.
I'm not particularly fond of DI frameworks (and neither are some of
the folks work on/near go-ipfs), but it seems unlikely that somebody
will rewrite the dependency wiring and lifecycle hooks of go-ipfs, and
add dynamic extension points, so this seems like a palatable
compromise.
There are also problems that we should clean up in how model the
go-ipfs app in fx, such as:
- We make extensive use of nested fx.Options, which fx itself
discourages because it "limits the user's ability to customize their
application". It should be easy to flatten these out into a single
[]fx.Option slice.
- We pass around a list of opaque libp2p opts, which makes it hard to
customize after-the-fact...we should consider naming each of these
opts and providing them to fx as proper dependencies, so that they can
be explicitly overridden.
- We call fx.Invoke() in some places with anonymous functions. We
should instead only pass exported functions to fx.Invoke(), so that
they have exported names, which would make it easier to remove/augment
the invocations that happen when the app is initialized.
These aren't blocking issues, they just make it harder and more
brittle to customize go-ipfs with this plugin.
Support hashes from 20 <= x <= 128 bytes, but does not yet have any option to generate anything else than 32 bytes hashes.
This will be forward compatible when we add such option.
The file contains the gateway your node is hosting in the http://<host>:<port> RFC 3986 format.
Structurally it works exactly the same as the API file.
This fixes a regression around directory listing and index.html hosting.
Seems that during one of recent refactors code changed and we no longer
check for trailing slash in HTTP request path, but look at content path
instead.
This cleans this up and also ensures dir behavior is the same for
both index.html hosting and dir-index-html (generated listing).
It also adds more tests so we catch any future regressions.
This fixes a regression introduced in 0.13.0, where websites hosted via
index.html placed in UnixFS directory were always returned with
Cache-Control: public, max-age=29030400, immutable
even when loaded from mutable /ipns/ contentPath.
* Delegated Routing.
Implementation of Reframe specs (https://github.com/ipfs/specs/blob/master/REFRAME.md) using go-delegated-routing library.
* Requested changes.
* Init using op string
* Separate possible ContentRouters for TopicDiscovery.
If we don't do this, we have a ciclic dependency creating TieredRouter.
Now we can create first all possible content routers, and after that,
create Routers.
* Set dht default routing type
* Add tests and remove uneeded code
* Add documentation.
* docs: Routing.Routers
* Requested changes.
Signed-off-by: Antonio Navarro Perez <antnavper@gmail.com>
* Add some documentation on new fx functions.
* Add changelog entry and integration tests
* test: sharness for 'dht' in 'routing' commands
Since 'routing' is currently the same as 'dht' (minus query command)
we need to test both, that way we won't have unnoticed divergence
in the default behavior.
* test(sharness): delegated routing via reframe URL
* Add more tests for delegated routing.
* If any put operation fails, the tiered router will fail.
* refactor: Routing.Routers: Parameters.Endpoint
As agreed in https://github.com/ipfs/kubo/pull/8997#issuecomment-1175684716
* Try to improve CHANGELOG entry.
* chore: update reframe spec link
* Update go-delegated-routing dependency
* Fix config error test
* use new changelog format
* Remove port conflict
* go mod tidy
* ProviderManyWrapper to ProviderMany
* Update docs/changelogs/v0.14.md
Co-authored-by: Adin Schmahmann <adin.schmahmann@gmail.com>
Co-authored-by: Marcin Rataj <lidel@lidel.org>
Co-authored-by: Adin Schmahmann <adin.schmahmann@gmail.com>
* feat: upgrade to go-libp2p-kad-dht@v0.16.0
* rename "cid format --codec" to "cid format --mc"
* refactor(test): from --codec to --mc
Co-authored-by: Marcin Rataj <lidel@lidel.org>
* feat: disable resource manager by default
We are disabling this by default for v0.13 as we work to improve the
UX around Resource Manager. It is still usable and can be enabled in
the IPFS config with "ipfs config --bool Swarm.ResourceMgr.Enabled true".
We intend to enable Resource Manager by default in a subsequent
release.
* docs(config): Swarm.ResourceMgr disabled by default
Co-authored-by: Marcin Rataj <lidel@lidel.org>
This fix safelists additional headers allowing JS running on websites to
read them when IPFS resource is downloaded via Fetch API.
These headers provide metadata necessary for making smart caching
decisions when IPFS resources are downloaded via Service Worker or a
similar middleware on the edge.
* Add 'ipfs repo migrate' command
This PR replaces #7658 that was originally contributed by zaibons, in order to move code into a branch and avoid some CI problem.
The command allows the user to run the repo migration without starting the daemon.
resolves#7471
* return non-ErrNeedMigration errors from fsrepo.Open()
Co-authored-by: Gus Eggert <gus@gus.dev>
* fix(core/gateway): option to limit directory size listing
* feat(gw): HTMLDirListingLimit
This is alternative take on the way we limit the HTML listing output.
Instead of a hard cut-off, we list up to HTMLDirListingLimit.
When a directory has more items than HTMLDirListingLimit we show
additional header and footer informing user that only $HTMLDirListingLimit
items are listed. This is a better UX.
* fix: 0 disables Gateway.HTMLDirListingLimit
* refactor: Gateway.FastDirIndexThreshold
see explainer in docs/config.md
* refactor: prealoc slices
* docs: Gateway.FastDirIndexThreshold
* refactor: core/corehttp/gateway_handler.go
https://github.com/ipfs/go-ipfs/pull/8853#discussion_r851437088
* docs: apply suggestions from code review
Co-authored-by: Alan Shaw <alan.shaw@protocol.ai>
Co-authored-by: Marcin Rataj <lidel@lidel.org>
Co-authored-by: Alan Shaw <alan.shaw@protocol.ai>
* update go-libp2p to v0.19.0
* chore: go-namesys v0.5.0
* refactor(config): cleanup relay handling
* docs(config): document updated defaults
* fix(tests): panic during sharness
* fix: t0160-resolve.sh
See https://github.com/ipfs/go-namesys/pull/32
* fix: t0182-circuit-relay.sh
* test: transport encryption
Old tests were no longer working because go-libp2p 0.19 removed
the undocumented 'ls' pseudoprotocol.
This replaces these tests with handshake attempt (name is echoed back on
OK or 'na' is returned when protocol is not available) for tls and noise
variants + adds explicit test that safeguards us against enabling
plaintext by default by a mistake.
* fix: ./t0182-circuit-relay.sh
test is flaky, for now we just restart the testbed when we get
NO_RESERVATION error
* refactor: AutoRelayFeeder with exp. backoff
It starts at feeding peers ever 15s, then backs off each time
until it is done once an hour
Should be acceptable until we have smarter mechanism in go-lib2p 0.20
* feat(AutoRelay): prioritize Peering.Peers
This ensures we feed trusted Peering.Peers in addition to any peers
discovered over DHT.
* docs(CHANGELOG): document breaking changes
Co-authored-by: Marcin Rataj <lidel@lidel.org>
Co-authored-by: Gus Eggert <gus@gus.dev>
* feat: persist limit changes to config
This changes the "ipfs swarm limit" command so that when limit changes
are applied via the command line, they are persisted to the repo
config, so that they remain in effect when the daemon restarts.
Any existing limit.json can be dropped into the IPFS config easily
using something like:
cat ~/.ipfs/config | jq ".Swarm.ResourceMgr.Limits = $(cat limit.json)" | sponge ~/.ipfs/config
This also upgrades to Resource Manager v0.3.0, which exports the config
schema so that we don't have to maintain our own copy of it.
Co-authored-by: Marcin Rataj <lidel@lidel.org>
* fix: go mod tidy
* fix: update expected prometheus metrics
Go 1.17 includes some new runtime metrics that are automatically
published by the Prometheus client, so this adds them to the expected
metric list so that the tests pass and we don't accidentally drop them
in the future.
Improves the way we handle If-None-Match header:
- Support for more than one Etag passed in If-None-Match
- Match both strong and weak Etags to maximize caching across
various HTTP clients and libraries (some send weak Etags by default)
- Support for wildcard '*'
- Tests for If-None-Match behavior
* Add initialization directory support to Docker image
* Add sharness test, fix bugs in init script
Fixed in init script:
- Added some missing quotes around expansions
- Fixed INIT_ARGS to not pass any args if IPFS_PROFILE isn't specified
- Use printf instead of "echo -e"
- Only run scripts in top-level of init dir
- Handle filenames correctly when finding init scripts (by using find + xargs)
* chore: docker cleanup
cleans up containers and images (useful when run on developer machine)
* remove container init documentation from README
There is already IPFS Docker documentation where this should live:
https://docs.ipfs.io/how-to/run-ipfs-inside-docker/
Co-authored-by: Caian <caian@ggaunicamp.com>
Co-authored-by: Marcin Rataj <lidel@lidel.org>
Co-authored-by: Gus Eggert <gus@gus.dev>
* feat: add block profiling to collect-profiles.sh
* feat: add more profiles to 'ipfs diag profile'
This adds mutex and block profiles, and brings the command up-to-par
with 'collect-profiles.sh', so that we can remove it.
Profiles are also now collected concurrently, which improves the
runtime from (profile_time * num_profiles) to just (profile_time).
Note that this has a backwards-incompatible change, removing
--cpu-profile-time in favor of the more general --profile-time, which
covers all sampling profiles.
* docs(cli): ipfs diag profile
* add CLI flag to select specific diag collectors
Co-authored-by: Marcin Rataj <lidel@lidel.org>
* update go-libp2p to v0.18.0
* initialize the resource manager
* add resource manager stats/limit commands
* load limit file when building resource manager
* log absent limit file
* write rcmgr to file when IPFS_DEBUG_RCMGR is set
* fix: mark swarm limit|stats as experimental
* feat(cfg): opt-in Swarm.ResourceMgr
This ensures we can safely test the resource manager without impacting
default behavior.
- Resource manager is disabled by default
- Default for Swarm.ResourceMgr.Enabled is false for now
- Swarm.ResourceMgr.Limits allows user to tweak limits per specific
scope in a way that is persisted across restarts
- 'ipfs swarm limit system' outputs human-readable json
- 'ipfs swarm limit system new-limits.json' sets new runtime limits
(but does not change Swarm.ResourceMgr.Limits in the config)
Conventions to make libp2p devs life easier:
- 'IPFS_RCMGR=1 ipfs daemon' overrides the config and enables resource manager
- 'limit.json' overrides implicit defaults from libp2p (if present)
* docs(config): small tweaks
* fix: skip libp2p.ResourceManager if disabled
This ensures 'ipfs swarm limit|stats' work only when enabled.
* fix: use NullResourceManager when disabled
This reverts commit b19f7c9eca.
after clarification feedback from
https://github.com/ipfs/go-ipfs/pull/8680#discussion_r841680182
* style: rename IPFS_RCMGR to LIBP2P_RCMGR
preexisting libp2p toggles use LIBP2P_ prefix
* test: Swarm.ResourceMgr
* fix: location of opt-in limit.json and rcmgr.json.gz
Places these files inside of IPFS_PATH
* Update docs/config.md
* feat: expose rcmgr metrics when enabled (#8785)
* add metrics for the resource manager
* export protocol and service name in Prometheus metrics
* fix: expose rcmgr metrics only when enabled
Co-authored-by: Marcin Rataj <lidel@lidel.org>
* refactor: rcmgr_metrics.go
* refactor: rcmgr_defaults.go
This file defines implicit limit defaults used when Swarm.ResourceMgr.Enabled
We keep vendored copy to ensure go-ipfs is not impacted when go-libp2p
decides to change defaults in any of the future releases.
* refactor: adjustedDefaultLimits
Cleans up the way we initialize defaults and adds a fix for case
when connection manager runs with high limits.
It also hides `Swarm.ResourceMgr.Limits` until we have a better
understanding what syntax makes sense.
* chore: cleanup after a review
* fix: restore go-ipld-prime v0.14.2
* fix: restore go-ds-flatfs v0.5.1
Co-authored-by: Lucas Molas <schomatis@gmail.com>
Co-authored-by: Marcin Rataj <lidel@lidel.org>
* feat: add support for carv2 import
* update to multicodec 0.4.0
* add sharness test for carv2
Co-authored-by: Keenan Nemetz <keenan.nemetz@gmail.com>
* 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>
* fix(gw): validate requested CAR version
This adds validation of 'application/vnd.ipld.car;version=n' passed
in the Accept header by HTTP clients to align Gateway behavior with
the spec submitted to IANA.
* test: fix comment in test/sharness/t0118-gateway-car.sh
Co-authored-by: Gus Eggert <gus@gus.dev>
Co-authored-by: Gus Eggert <gus@gus.dev>
The Docker sharness tests were disabled years ago when go-ipfs moved
from Travis to CircleCI. This makes the tweaks necessary to re-enable
them.
The Docker image has since moved to be based on BusyBox which doesn't
have the requisite wget version for the existing tests to work, so
this adds some functionality to the pollEndpoint program to support
polling HTTP endpoints as well.
