-
-
Notifications
You must be signed in to change notification settings - Fork 3.1k
Description
go-ipfs 0.11.0 Release Notes
We're happy to announce go-ipfs 0.11.0. This release comes with improvements to the UnixFS Sharding and PubSub experiments as well as support for Circuit-Relay v2 which sets the network up for decentralized hole punching support.
As usual, this release includes important fixes, some of which may be critical for security. Unless the fix addresses a bug being exploited in the wild, the fix will not be called out in the release notes. Please make sure to update ASAP. See our release process for details.
🛠 BREAKING CHANGES
- UnixFS sharding is now automatic and enabled by default
- HAMT-based sharding is applied to large directories (i.e. those that would serialize into block larger than ~256KiB)s. This means importing data via commands like
ipfs add -r <directory>
may result in different CIDs due to the different DAG representations. - Support for
Experimental.ShardingEnabled
is removed.
- HAMT-based sharding is applied to large directories (i.e. those that would serialize into block larger than ~256KiB)s. This means importing data via commands like
- go-ipfs can no longer act as a Circuit Relay v1
- Node will refuse to start if
Swarm.EnableRelayHop
is set totrue
- If you depend on v1 relay service provider, see "Removal of v1 relay service" section for available migration options.
- Node will refuse to start if
- HTTP RPC wire format for experimental commands at
/api/v0/pubsub
changed.- If you use js-ipfs-http-client or go-ipfs-http-client, just update to their latest version.
- If you use something else, see "Multibase in PubSub" section below for migration details.
Keep reading to learn more details.
🔦 Highlights
🗃️ Automatic UnixFS sharding
Truly big directories can have so many items, that the root block with all of their names is too big to be exchanged with other peers. This was partially solved by HAMT-sharding, which was introduced a while ago as opt-in. The main downside of the implementation was that it was a global flag that sharded all imported directories (big and small).
This release solves that inconvenience by making UnixFS sharding smarter and applies it only to larger directories (i.e. directories that would be at least ~256KiB). This is now the default behavior in ipfs add
and ipfs files
commands, where UnixFS sharding works out-of-the-box.
🔁 Circuit Relay v2
This release adds support for the circuit relay v2 protocol based on the reference implementation from go-libp2p 0.16.
This is the cornerstone for maximizing p2p connections between IPFS peers. Every publicly dialable peer can now act as a limited relay v2, which can be used for hole punching and other decentralized signaling protocols.
Limited relay v2 configuration options
go-ipfs can now be configured to act as a RelayClient
that uses other peers for autorelay functionality when behind a NAT, or provide a limited RelayService
to other peers on the network.
Starting with go-ipfs v0.11 every publicly dialable go-ipfs (based on AutoNAT determination) will start a limited RelayService
. RelayClient
remains disabled by default for now, as we want the network to update and get enough v2 service providers first.
Note: the limited Circuit Relay v2 provided with this release only allows low-bandwidth protocols (identify, ping, holepunch) over transient connections. If you want to relay things like bitswap sessions, you need to set up a v1 relay by some other means. See details below.
Removal of unlimited v1 relay service provider
Switching to v2 of the relay spec means removal or deprecation of configuration keys that were specific to v1.
- Relay transport and client support circuit-relay v2:
Swarm.EnableAutoRelay
was replaced bySwarm.RelayClient.Enable
.Swarm.DisableRelay
is deprecated, relay transport can be now disabled globally (both client and service) by settingSwarm.Transports.Network.Relay
tofalse
- Relay v1 service provider was replaced by v2:
Swarm.EnableRelayHop
no longer starts an unlimited v1 relay. If you have it set totrue
the node will refuse to start and display an error message.- Existing users who choose to continue running a v1 relay should migrate their setups to relay v1 based on js-ipfs running in node, or the standalone libp2p-relay-daemon configured with
RelayV1.Enabled
set totrue
. Be mindful that v1 relays are unlimited, and one may want to set up some ACL based either on PeerIDs or Subnets.
🕳 Decentralized Hole Punching (DCUtR protocol client)
We are working towards enabling hole punching for NAT traversal when port forwarding is not possible.
go-libp2p 0.16 provides an implementation of the DCUtR (decentralized hole punching) protocol. It is hidden behind the Swarm.EnableHolePunching
configuration flag.
When enabled, go-ipfs will coordinate with the counterparty using a relayed v2 connection, to upgrade to a direct connection through a NAT/firewall whenever possible.
This feature is disabled by default in this release, but we hope to enable it by default as soon the network updates to go-ipfs v0.11 and gains a healthy set of limited v2 relays.
💬 Multibase in PubSub HTTP RPC API
This release fixed some edge cases that were reported by users of the PubSub experiment, getting it closer to becoming a stable feature of go-ipfs. Some PubSub users will notice that the plaintext limitation is lifted: one can now use line breaks in messages published to non-ascii topic names, or even publish arbitrary bytes to arbitrary topics. It required a change to the wire format used when pubsub commands are executed over the HTTP RPC API at /api/v0/pubsub/*
, and also modified the behavior of the ipfs pubsub pub
command, which now is publishing only a single pubsub message with data read from a file or stdin.
PubSub client migration tips
If you use the HTTP RPC API with the go-ipfs-http-client library, make sure to update to the latest version. The next version of js-ipfs-http-client will use the new wire format as well, so you don't need to do anything.
If you use /api/v0/pubsub/*
directly or maintain your own client library, you must adjust your HTTP client code. Byte fields and URL args are now encoded in base64url
Multibase. Encode/decode bytes using the ipfs multibase --help
commands, or use the multiformats libraries (js-multiformats, go-multibase).
Low level changes:
topic
passed as URLarg
in requests to/api/v0/pubsub/*
must be encoded in URL-safe multibase (base64url
)data
,from
,seqno
andtopicIDs
returned in JSON responses are now encoded in multibase- Peer IDs returned in
from
now use the same default text representation from go-libp2p and peerid encoder/decoder from libp2p. This means the same text representation as in as inswarm peers
, which makes it possible to compare them without decoding multibase. /api/v0/pubsub/pub
no longer acceptsdata
to be passed as URL, it has to be sent asmultipart/form-data
. This removes size limitations based on URL length, and enables regular HTTP clients to publish data to PubSub topics. For example, to publishsome-file
to topic namedtest-topic
using vanillacurl
, one would execute:curl -X POST -v -F "stdin=@some-file" 'http://127.0.0.1:5001/api/v0/pubsub/pub?arg=$(echo -n test-topic | ipfs multibase encode -b base64url)'
ipfs pubsub pub
on the command line no longer accepts variadicdata
arguments. Instead, it expects a single file input or stream of bytes from stdin. This ensures arbitrary stream of bytes can be published, removing limitation around messages that include\n
or\r\n
.
⚙️ New configuration flags
Addresses.AppendAnnounce
is an array of multiaddrs, similar toAddresses.Announce
, except it does not override inferred swarm addresses, but appends custom ones to the list.- Pubsub experiments can now be enabled via config, removing the need for CLI flag to be passed every time daemon starts:
Pubsub.Enabled
enables the pubsub system.Ipns.UsePubsub
enables IPFS over pubsub experiment for publishing IPNS records in real time.
🔐 Support for DAG-JOSE IPLD codec
JOSE is a standard for signing and encrypting JSON objects. DAG-JOSE is an IPLD codec based on JOSE and represented in CBOR. Upon encountering the dag-jose
multicodec indicator, implementations can expect that the block contains dag-cbor encoded data which matches the IPLD schema from the DAG-JOSE spec.
This work was contributed by Ceramic and acts as a template for future IPFS improvements driven by the real world needs of the IPFS community.
🗺 What's left for release
- Tracking issue for UnixFS automatic sharding #8106
- Pubsub subscription data encoding #7939 (fix: multibase in pubsub http rpc #8183)
- context/tracing in the blockstore/datastore pipeline #6803
- WIP: update go-libp2p to v0.16.0 #8522
- feat: enabling pubsub and ipns-pubsub via config flags #8510
- feat: Addresses.AppendAnnounce #8177
- fix: multiple subdomain gateways on same domain #8556
- refactor: remove dir-index-html submodule #8555
- fix(unixfs): check for errors before dereferencing the link #8508
- chore: replace go-merkledag walk with go-ipld-prime traversal for dag export #8506
- Automate copying of signed binaries from dist.ipfs.io/go-ipfs #8316
- api/v0/get?archive=true: tar: time stamp is in the future #8406
- Multipart reader regression in go 1.17 #8445
- Automated publishing to Chocolatey #8252
- Move Docker build to Actions #8330
- Support dag-jose codec by default #8364
- chore: update deps #8571
🚢 Estimated shipping date
RC1: 2021-11-29
ECD: 2021-12-06
✅ Release Checklist
For each RC published in each stage:
- version string in
version.go
has been updated (in therelease-vX.Y.Z
branch). - tag commit with
vX.Y.Z-rcN
- upload to dist.ipfs.io
- Build: https://github.com/ipfs/distributions#usage.
- Pin the resulting release.
- Make a PR against ipfs/distributions with the updated versions, including the new hash in the PR comment.
- Ask the infra team to update the DNSLink record for dist.ipfs.io to point to the new distribution.
- cut a pre-release on github and upload the result of the ipfs/distributions build in the previous step.
- Announce the RC:
- On Matrix (both #ipfs and #ipfs-dev)
- To the early testers listed in docs/EARLY_TESTERS.md. Do this by copy/pasting their GitHub usernames and checkboxes as a comment so they get a GitHub notification. (example)
Checklist:
- Stage 0 - Automated Testing
- Fork a new branch (
release-vX.Y.Z
) frommaster
and make any further release related changes to this branch. If any "non-trivial" changes (see the footnotes of docs/releases.md for a definition) get added to the release, uncheck all the checkboxes and return to this stage.- Follow the RC release process to cut the first RC.
- Bump the version in
version.go
in themaster
branch tovX.(Y+1).0-dev
.
- Automated Testing (already tested in CI) - Ensure that all tests are passing, this includes:
- unit, sharness, cross-build, etc (
make test
) - lint (
make test_go_lint
) - interop
- go-ipfs-api
- go-ipfs-http-client
- WebUI
- unit, sharness, cross-build, etc (
- Fork a new branch (
- Stage 1 - Internal Testing
- CHANGELOG.md has been updated
- use
./bin/mkreleaselog
to generate a nice starter list
- use
- Infrastructure Testing:
- Deploy new version to a subset of Bootstrappers
- Deploy new version to a subset of Gateways
- Deploy new version to a subset of Preload nodes
- Collect metrics every day. Work with the Infrastructure team to learn of any hiccup
- IPFS Application Testing - Run the tests of the following applications:
- IPFS Desktop
- Ensure the RC is published to the NPM package (happens automatically, just wait for CI)
- Upgrade to the RC in ipfs-desktop and push to a branch (example), and open a draft PR to track through the final release (example)
- Ensure CI tests pass, repeat for new RCs
- IPFS Companion - @lidel
- IPFS Desktop
- CHANGELOG.md has been updated
- Stage 2 - Community Prod Testing
- Documentation
- Ensure that CHANGELOG.md is up to date
- Ensure that README.md is up to date
- Update docs by merging the auto-created PR in https://github.com/ipfs/ipfs-docs/pulls (they are auto-created every 12 hours)
- Invite the wider community through (link to the release issue):
- discuss.ipfs.io
- IRC
- Documentation
- Stage 3 - Release
- Final preparation
- Verify that version string in
version.go
has been updated. - Merge
release-vX.Y.Z
into therelease
branch. - Tag this merge commit (on the
release
branch) withvX.Y.Z
. - Release published
- to dist.ipfs.io
- to npm-go-ipfs
- to chocolatey
- to snap
- to github
- use the artifacts built in CI for dist.ipfs.io
- to arch (flag it out of date)
- Cut a new ipfs-desktop release
- Verify that version string in
- Submit this form to publish a blog post, linking to the GitHub release notes
- Broadcasting (link to blog post)
- Twitter (request in Slack channel #pl-marketing-requests)
- Matrix
- discuss.ipfs.io
- Final preparation
- Post-Release
- Merge the
release
branch back intomaster
, ignoring the changes toversion.go
(keep the-dev
version from master). - Create an issue using this release issue template for the next release.
- Make sure any last-minute changelog updates from the blog post make it back into the CHANGELOG.
- Mark PR draft created for IPFS Desktop as ready for review.
- https://github.com/ipfs/go-ds-s3/pull/210/files
- Deploy our final release to our infra
- Merge the
⁉️ Do you have questions?
The best place to ask your questions about IPFS, how it works and what you can do with it is at discuss.ipfs.io. We are also available at the #lobby:ipfs.io
Matrix channel which is bridged with other chat platforms.
Release improvements for next time
< Add any release improvements that were observed this cycle here so they can get incorporated into future releases. >