fix(ci): retry nargo dep + solc downloads to survive transient DNS drops#23490
Merged
Conversation
alexghr
approved these changes
May 22, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Why
Merge-train/spartan keeps failing on transient DNS resolution errors, e.g.:
7 such failures in the last week (github.com ×3 via nargo, binaries.soliditylang.org ×1 via solc; release-assets.githubusercontent.com ×3 already fixed by #23333). Root cause is almost certainly the EC2 VPC resolver's ~1024 packets/sec-per-ENI cap being exhausted by heavy parallel builds, so lookups are silently dropped.
This is the cheap, immediate mitigation: retry the two un-retried network fetches that bite the merge train. It does not fix the root cause — a host-local caching resolver does (dnsmasq spike linked below, for the future).
What
noir-projects/bootstrap.sh— wrap the nargo dependency-download (fmt --checkprep step) inci3/retry. On failure it wipes the partial dependency cache ($HOME/nargo) before retrying: a half-finished clone is exactly what produces theCannot read file .../Nargo.tomlerror, so a naive retry would just re-hit the poisoned dir. A warm cache is left intact on success.l1-contracts/bootstrap.sh— wrap theforge build --usesvm solc download inci3/retry. The merge queue disables the S3 cache, so this download path runs on every merge-train build.Both use the existing
ci3/retryhelper (3 attempts;RETRY_SLEEP=10for the nargo step to give DNS a little longer to recover).Dropped
An earlier commit hardened the runner's
/etc/resolv.conf(options timeout:1 attempts:5 rotate+ a public fallback nameserver). Removed —rotateto a public resolver is brittle and risks breaking any VPC-private name resolution. Not worth the blast radius for a mitigation.Future: root-cause fix
Host-local dnsmasq caching resolver on the runner — what it would look like: https://gist.github.com/AztecBot/a22cc18bd30ec0bd3dff72b70d675304