Cobra Forum

Question: Potential Nginx Configuration Issues After Updating to Plesk Obsidian 18.0.72 Update 2 on Specific Virtualized Setups


Applicable to:

• Plesk for Linux

Symptoms

• After updating Plesk to Obsidian 18.0.72 Update 2, all domains nginx configuration files are broken.
• Domains show a http 404 error.
• AlmaLinux OS is in use.
• IPv6 addresses are used on the server.
• The CPU with 32-bit cache size or QEMU Virtual CPU is used.

Cause

This behavior was identified as bug PPPM-15110. In some virtualization setups that use QEMU-based virtual CPUs, Nginx can incorrectly detect the CPU cache line size. This misdetection often results in a default map_hash_bucket_size of 34, which may lead to configuration issues, particularly when processing long IPv6 addresses.


Resolution


Until the bug is fixed, refer to the following workaround to increase the map_hash_bucket_size parameter:

1.Connect to the server via SSH

2.Create a configuration file with the parameter added:

# echo 'map_hash_bucket_size 64;' >> /etc/nginx/conf.d/fix_hash_bucket.conf
3.Restart nginx:

# systemctl restart nginx
4.Rebuild the web server configuration:

# plesk sbin httpdmng --reconfigure-all