Skip to content

ci(testnet): pruned-storage scenarios - #23

Merged
EnzinoBB merged 7 commits into
masterfrom
testnet/pruned
Oct 11, 2026
Merged

EnzinoBB merged 7 commits into
masterfrom
testnet/pruned

Conversation

@EnzinoBB

Copy link
Copy Markdown
Collaborator

What this adds

Test-network scenarios for pruned storage (#22). Both are off by default.

Input / env What happens
prune_node / PRUNE_NODE That node runs with a checkpoint every 1,000 blocks, checkpoint_keep=2 and prune_keep_blocks=300. Combined with restart_at, it is restarted from its pruned storage.
join_pruned_at / JOIN_PRUNED_AT One more node starts once the network reaches that sequence. It is in nobody's hosts or trusted list and has no funds or checkpoint. It syncs every block from block 0 through the normal block requests and prunes behind itself.

The run fails if a pruned node never prunes. It reports each pruned node's pruning log lines and its block DB file by file, next to a full node's. As for every node, the state digests of the pruned nodes must match the others'.

Runs (with #22)

  • 38066244096: n7 pruned below 700, 1,700 and 2,700, restarted from its pruned storage, and matched the digests to block 4,200.
  • 38069311587: n7's blockchain.db is 1.6 MB, against 4.4 MB on a full node.
  • 38073551101: n8 joins at block 1,500 and matches all 8 nodes' digests to block 4,200.
  • 38076463340: n8 joins at block 3,507 and prunes below 700 and 1,700 while catching up. Its blockchain.db is 2.0 MB against 6.3 MB, and all 8 nodes' digests match to block 4,600.

🤖 Generated with Claude Code

EnzinoBB and others added 7 commits October 10, 2026 15:55
With PRUNE_NODE (workflow input prune_node) that node runs with a checkpoint every 1'000 blocks,
checkpoint_keep=2 and prune_keep_blocks=300. The run fails if it never prunes, and reports the
pruning log lines and every node's block DB size; its state digest is checked as for any node.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
With JOIN_PRUNED_AT (workflow input join_pruned_at) one more node, in nobody's hosts or trusted
list and without funds or checkpoints, starts once the network reaches that sequence. It syncs
every block from block 0 through the normal block requests, validates and applies them, saves a
checkpoint every 1'000 blocks and prunes behind itself (prune_keep_blocks=300). The run waits for
it to reach the target too, requires it to have pruned, and checks its state digests with the
others'.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A node that dies is now reported with its exit status (and signal), and with core dumps on
(kernel.core_pattern into the run directory, ulimit -c unlimited) the run prints the backtrace
of every core file with gdb.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@EnzinoBB
EnzinoBB merged commit 809b8ed into master Oct 11, 2026
2 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant