Overview

A bug in rXg versions between 15.435 and 15.459 causes the yarn.lock file to grow exponentially each time rxgd starts. Eventually the file becomes too large for yarn to process, resulting in hangs during rxgd startup.

Problem / Question

• rxgd hangs on startup
• "File too large" or bundle errors in logs
• /space/rxg/console/yarn.lock is abnormally large

Root Cause

A bug causes the yarn.lock file to grow exponentially with each rxgd restart. Fixed in version 15.459 and later.

Resolution

Single Node

1. Replace the corrupted yarn.lock:

   wget https://build.rgnets.com/build/patches/yarn.lock -O /space/rxg/console/yarn.lock
   

2. Upgrade to 15.459 or higher

All Nodes in a Cluster

Run on every node:

cd /space/rxg && curl https://build.rgnets.com/build/patches/2902.patch | git apply
wget https://build.rgnets.com/build/patches/yarn.lock -O /space/rxg/console/yarn.lock
rr

If Already on 15.465 or Later

The yarn.lock bug is fixed in 15.459+, so if on 15.465 the symptoms are likely caused by something else. One known case involved a lock on the cluster database.

Diagnosis: Check tfpostgres for lock-related messages:

DETAIL:  Processes holding the lock:

Resolution: Identify and kill the blocking process. In one case it was a purge database process trying to drop chunks from the SUL stats table.