Skip to content

Latest commit

 

History

History
101 lines (80 loc) · 5.95 KB

RELEASE-NOTES-5.6.0.md

File metadata and controls

101 lines (80 loc) · 5.95 KB
Raw

About this release

This is a maintenance Lima release.

It:

  • Enchances the p2p noise protocol with an optional setting to ask a peer for the version of their node. A node operator can set one's node to not respond to this message if one considers this a private information. This is to be used for monitoring the network's health.

  • Enhances FSM behaviour: when the initiator is offline, allows the responder to stay online waiting for it even if the timeout timer is reached.

  • Configuration values can now be set using OS environment variables, where the environment variable name is on the form AE__k1__k2, e.g. AE__CHAIN__PERSIST=true. Note that two underscores are used to separate each level. Structured values must be JSON-encoded. The prefix AE can not be customized. The environment variables are applied after reading an available config file, and all values are checked against the schema. See docs/configuration.md.

  • Introduces persistence of the peer pool: if the node flag for disc persistence is set, after a node restart old peers are loaded from the DB. Trusted peers are provided to the node from the config file so they are already persisted. Since the time being off is unknown, all peers are loaded as unverified, even if they had been verified before the node stop.

  • Changes the default value and significantly speeds up the TCP probes. Those used to be done roughly once every 2 minutes while now they are executed every second. Note that this is done only if there are peers to be checked: if a peer is missing, there is a grace period in which we mark it as 'suspended', waiting for its return. If all peers are checked, we wait longer - until either a new peer is added or there is a not-suspended peer. Basically the probes come on demand. Even despite the probes are fast, there is a maximum size of currently ongoing probes.

  • Fixes the process of probing verified peers. Until now we were probing only unverified peers. If a peer responds to the probe - we consider it verified. We downgrade a verified peer only after we fail connecting to it. This resulted in amounting peers that used to be verified at one point of time. This is fixed now as we probe verified peers as well. What is more is, that this PR plays really well with PR #3365. This means that if a peer goes down, we will find it significantly sooner and will downgrade it as an unverified one. This is important as a nodes shares to other nodes only peers one considers to be verified.

  • Added support for reporting chain events during contract evaluation. These events are published internally as 'tx_events' (available e.g. for plugins), and can also be fetched via the HTTP 'dry-run' method. The events are presented as 'dummy' transactions, unsigned, reflecting the details of the respective events.

  • Fixes a bug: there are special trusted peers that are never deleted nor downgraded. Those are quite unique as the node consideres them to be available. One can pick their own set of trusted peers but this was not really effective: the default peers were always included as a bonus. This PR fixes it: if the new config flag is set to false - the defaults are omited. The flag is called include_default_peers and by default it is true.

  • There used to be multiple copies of the same peers in peers pool. This resulted in propagation of same peers with different Peer IDs through the network. Now duplicates are cleaned up.

  • Some further peers fine tuning: we used to treat peers that refuse connecting to us as they were malicious. This was the case no matter if peers were active and part of the network and only simply refusing to open a enoise connection to our node (most likely because their inbound pool is staturated already). This is now changed and if the node is responding but refusing to open an enoise connection - we consider it still a valid and verified participant of the network.

  • Improves sync's peer propagation: peers are split into verified and unverified. We had allowed a grace period for peers to be down while the node still considers them as verified and broadcasts them as such. Once a verified peer is being detected as down, it enters a standby state that allows them to stay verified. This PR tightens the process - once a peer is being detected to be unreachable - it is being downgraded from verified to unverified immediately. The processing of unverified being deleted is left unchanged - they still enter the standby mode for a few times before being removed.

  • Fixes inconsistency of the /status endpoint

  • Provides some fine tuning to the supervision tree of the conductor. This will prevent the node making too many recovery attempts after a crash.

  • Enhances FSM behaviour: when the initiator is offline, allows the responder to stay online waiting for it even if the timeout timer is reached.

  • Adds an explicit settings for local_lima_testnet. So far those were implicit as Lima is the currently running release. The MDW though uses a macro to parse those and if there is no explicit function head - it fails. This aims at providing the MDW with capabilities to work with a local Lima test environment.

  • Due to MacOS High Sierra reaching EOL status we no longer build and provide release binaries for MacOS High Sierra. MacOS Mojave and above is supported.

Please join the mainnet by following the instructions in the documentation below, and let us know if you have any problems by opening a ticket. Troubleshooting of common issues is documented in the wiki.

Documentation

For an overview of the installation process for different platforms, building the package from source, configuration and operation of the Aeternity node please refer to Aeternity node documentation.