Edge packet drops due to the Edge being busy trying to write large files to frr log as the the log rotation for frr failed in the Edge nodes

Products

• VMware NSX

Issue/Introduction

• Running NSX version 4.2.1
• The Edge node /var/log/ partition grows in size.
• If you log in as root and list the the files in /var/log/, you noticed that all /var/log files are growing quite large. 
  
  
   
• The /var/log/frr file path shows only frr.log growing huge almost 28GB and no rotation of frr logs are happening:

         

• Due to this disk space usage, Edge node log extraction will be failing on the NSX UI.
• You may notice intermittent packet drops for traffic traversing the edge node.

Environment

VMware NSX 4.2.1

Cause

This is a known issue impacting VMware NSX 4.2.1.

When the NSX upgrade happens to 4.2.1, the frr file is removed from the folder /etc/logrotate.d causing the frr log rotation to stop working.

Note: /etc/logrotate.d/frr does not exist and the issue is seen on fresh deployments of NSX 4.2.1.

Resolution

The issue has been fixed in NSX 4.2.1.3.

Workaround:

Copy and paste the content of frr file (attached to this KB) in the /etc/logrotate.d using vi/vim editor with file name as frr and this should start the frr log rotation automatically every hour