Repository navigation
docs(node): retire the random peer list in favour of the Sei Labs seeds - #96
Conversation
The random peer snippet shipped 17 hardcoded community addresses, and only 6 of them still answer: 2 refuse the connection outright and 9 time out. Operators following the node guide were copying a list that is roughly two thirds dead, which looks like a broken network rather than a stale document. They no longer need it. Since v6.6.2, `seid init` writes the Sei Labs seeds into bootstrap-peers for well-known public networks, so peer discovery works with no manual configuration. index.mdx already said so in the init block while still offering the dead list above it, alongside a persistent-peers snippet that is the wrong field for seeds. Those are now one coherent section. Replaces the snippet with one that serves the seeds, in the same shape, so pages keep their copy button. The addresses match app/seeds in sei-chain v6.7.1 exactly, and all six were verified to complete a P2P handshake. Also corrects max-packet-msg-payload-size in the config.toml example, which read 10240 against an actual default of 1000000. llms-full.txt still carries the old snippet, deliberately left alone: AGENTS.md says not to hand-edit it, and the weekly regeneration workflow picks it up from the deployed site.
|
Preview deployment for your docs. Learn more about Mintlify Previews.
💡 Tip: Enable Automations to automatically generate PRs for you. |
PR SummaryLow Risk Overview
Troubleshooting now tells operators to reset to those seeds for handshake/ID mismatches, with a note that re- Technical reference fixes the Reviewed by Cursor Bugbot for commit 8ecac82. Bugbot is set up for automated code reviews on this repo. Configure here. |
There was a problem hiding this comment.
This PR swaps the mostly-dead community peer list for the Sei Labs seeds, fixes the persistent-peers vs bootstrap-peers mix-up, and corrects the max-packet-msg-payload-size default. It is clean overall, with only small non-blocking notes.
Findings: 0 blocking | 7 non-blocking | 3 posted inline
Blockers
- None at the file/PR level.
Non-blocking
- The Cursor second-opinion review produced no output. The Codex review reported no material issues.
- REVIEW_GUIDELINES.md is empty, so this review applies only AGENTS.md conventions.
- I found no remaining references to
random-peers.jsx/RandomPeersoutside the generatedllms-full.txt. Leaving that file for the weekly regeneration matches AGENTS.md. - The
format='bash'branch, which outputsSEEDS="...", is no longer used by any caller. Consider dropping it, or makingtomlthe default, so the snippet doesn't suggest aSEEDSshell variable that nothing reads. - 3 suggestion(s)/nit(s) flagged inline on specific lines.
| addresses, so a stale or mismatched node ID is the usual cause of this error: | ||
|
|
||
| <RandomPeers network="mainnet" format="toml" /> | ||
| <Seeds network="mainnet" format="toml" /> |
There was a problem hiding this comment.
[suggestion] This step only shows mainnet seeds. A Sei Testnet operator who hits peer handshake failed would copy pacific-1 seeds into an atlantic-2 node. Consider adding the testnet block with labels, as node/index.mdx does: - Sei Mainnet (pacific-1): / - Sei Testnet (atlantic-2):.
| className="inline-flex items-center gap-1 px-2 py-1 rounded-md text-[11px] font-semibold border border-neutral-300 dark:border-neutral-600 bg-white/80 dark:bg-neutral-800/80 text-neutral-700 dark:text-neutral-200 hover:bg-white dark:hover:bg-neutral-800 transition-colors"> | ||
| {copied ? ( | ||
| <> | ||
| <CheckIcon className="text-green-600" /> |
There was a problem hiding this comment.
[nit] The old snippet used text-green-600 dark:text-green-400 for the Copied check icon. The dark: variant was dropped here, so the icon has lower contrast in dark mode. Consider restoring it.
| <pre | ||
| className="m-0 p-3 rounded-md bg-neutral-100 dark:bg-neutral-800 text-sm opacity-70" | ||
| style={{ fontFamily: 'var(--sei-font-mono)' }}> | ||
| No seeds configured for network “{network}”. |
There was a problem hiding this comment.
[nit] This line uses curly quotes (“ ”), while the rest of the codebase uses straight quotes. Purely cosmetic.
Problem
snippets/random-peers.jsxserves 17 hardcoded community peer addresses, and most of them are gone. Dialling every entry today:That snippet is rendered on the node setup guide and the troubleshooting guide, so an operator following either one is copying a list that is roughly two thirds dead. It reads as a broken network rather than a stale page, and the troubleshooting guide in particular offers it as the fix for
peer handshake failed, which is the error a dead peer produces.There is also a contradiction on the setup page. It tells you peer discovery needs no manual configuration, then immediately offers the dead list above it, plus a
sedsnippet writing them topersistent-peers, which is the wrong field for seeds.Change
Operators do not need a peer list any more. Since v6.6.2,
seid initwrites the Sei Labs seeds intobootstrap-peersfor well-known public networks, so a fresh node bootstraps with no further configuration.random-peers.jsxwithseeds.jsx, serving the three Sei Labs seeds per network in the same shape, so both pages keep their copy buttonpersistent-peerssnippetinitwill not overwrite abootstrap-peersvalue you have already setmax-packet-msg-payload-sizein theconfig.tomlexample, which read10240against an actual default of1000000Verification
The six addresses match
app/seeds/seeds.goin sei-chain v6.7.1 exactly, and all six complete a P2P handshake:Other
[p2p]values in the example were checked againstDefaultP2PConfig()and theconfig.tomltemplate at v6.7.1 and are correct as written,max-outbound-connectionsincluded.Notes
llms-full.txtstill contains the old snippet. Left alone deliberately:AGENTS.mdsays not to hand-edit the generated.txtfiles, and the weeklyregenerate-llms.ymlworkflow rebuilds them from the deployed site.Supersedes #57, which predates the seed documentation already on
mainand had fallen into conflict.