OpenClaw 2026.5.12 OOM regression / gateway heap leak
A regression in OpenClaw 2026.5.12 OOMs on fresh CLI commands like gateway status and doctor, and a gateway heap leak grows to about 3GB in seven minutes and crash-loops - one report logged 50 restarts in eight hours. Downgrade with npm install -g openclaw@2026.3.11; raising NODE_OPTIONS max-old-space-size only delays the leak. Pin the older version until it's patched.
Likely cause
A regression in OpenClaw 2026.5.12 OOMs on fresh CLI commands like gateway status and doctor, plus a gateway heap leak that grows to about 3GB in roughly 7 minutes and crash-loops (one report logged 50 restarts in 8 hours).
The fix
- 1 Downgrade until patched: npm uninstall -g openclaw && npm install -g openclaw@2026.3.11.
- 2 NODE_OPTIONS=--max-old-space-size=4096 only delays the leak - it is not a fix.
- 3 Pin the older version in CI so an auto-update doesn't reintroduce it.
npm install -g openclaw@2026.3.11Hit a different error?
Paste any agent error and get the cause and fix in seconds.
Frequently asked questions
Even 'openclaw gateway status' OOMs now. Is my machine out of RAM?
Not necessarily. The 2026.5.12 regression OOMs on fresh CLI commands like status and doctor regardless of available memory. Downgrading to 2026.3.11 restores normal behavior.
Will raising the Node heap limit fix the crash loop?
No. NODE_OPTIONS=--max-old-space-size=4096 only delays the leak - the heap still grows to the new ceiling and OOMs. Downgrade and pin the older version until the leak is patched.
How do I stop an auto-update from bringing the bug back?
Pin openclaw@2026.3.11 explicitly in your install or CI step rather than installing latest, so a later auto-update can't silently reintroduce the regression.
Stop firefighting agent errors
Decoding errors one at a time is the manual version of what BetterClaw automates. Run your OpenClaw agents hosted with managed models, retries and config validation built in.
$19/month per agent · BYOK · 7-day money-back guarantee