arweave.net sandbox subdomain resolution broken after legacynet sunset — 502 globally

[AlexTheWizardL]

Description

Since Feb 24, 2026 (~10:30 UTC), all sandbox (hash-based) ArNS subdomain requests on arweave.net are returning 502 Bad Gateway or timing out. This started the day after the AO legacynet sunset (Feb 23), where a whitelist was deployed for old processes on Forward Research nodes.

Why this may be AO-related

ArNS sandbox subdomain resolution depends on AO — gateways fetch registry state from AO Compute Units to resolve <hash>.arweave.net paths. The failure pattern matches an AO CU dependency issue:

Endpoint	Works?	Requires AO?
arweave.net/graphql	Yes	No
arweave.net/raw/<txId>	Yes	No
/ar-io/healthcheck	Yes	No
Named ArNS (ardrive.arweave.net)	Yes	Cached/pre-resolved
Sandbox ArNS (<hash>.arweave.net/<txId>)	No	Yes

Everything that doesn't need AO resolution works. Everything that does is broken.

Repro

# Works — no AO resolution needed
curl -s -o /dev/null -w "%{http_code}\n" "https://arweave.net/raw/3DHmCQon0aYlaBKomz0y8VRdnw0hMqYgPNx4CboPsSs"
# 200

# Broken — requires sandbox subdomain resolution via AO CU
curl -s -o /dev/null -w "%{http_code}\n" -L "https://arweave.net/3DHmCQon0aYlaBKomz0y8VRdnw0hMqYgPNx4CboPsSs"
# 502

# Healthcheck on the broken subdomain reports OK
curl -s "https://3qy6mcike7i2mjlickujwpjs6fkf3hyneezkmib43r4atoqpwevq.arweave.net/ar-io/healthcheck"
# {"status":"ok"}

CDN77 returns: "CDN77 server is unable to reach website's origin server"

Impact

All permaweb apps accessed via arweave.net/<txId> (manifest deployments) are broken globally. Confirmed from Ukraine, Germany, and USA. Even the official ArweaveApps site (arweaveapps.com) fails when making requests through sandbox subdomains.

Example of a broken deployed app:
https://arweave.net/TYft-X7b7EC9JN5qCTeLJhUAoCfQhp0o32DLarNXGoQ/#/projects/9bbf901d-f378-493f-a2bb-fbc0aaf51e3c
https://r32tnk2qwp4a5dwluoy645gi52q6l4vw3wtwnsxf2q6lhjk2ngga.arweave.net/jvU2q1Cz-A6Oy6Ox7nTI7qHl8rbdp2bK5dQ8s6VaaYw/#/projects/9bbf901d-f378-493f-a2bb-fbc0aaf51e3c

Questions

1. Was the ArNS registry AO process (or the CUs that arweave.net depends on for resolution) affected by the legacynet whitelist? Does this process need to be migrated to mainnet?
2. Is there anything app developers need to do on their side to restore functionality — redeploy, update gateway config, migrate to a different AO process — or is this purely an infrastructure fix?
3. How does this affect apps that use a Vite proxy pointing to arweave.net for manifest resolution? Will the proxy continue to work once this is resolved, or do we need to change our setup?

Related

• ar-io issue: ArNS subdomain data requests returning 502/timeout while healthcheck reports OK ar-io/ar-io-node#619
• AO legacynet sunset: https://x.com/aoTheComputer/status/1891893166242422838
• ar.io CU setup context: https://x.com/dtfiedler/status/1889045690866872762