RabbitMQ's pattern_to_regex function maps SQL-like patterns to regular expressions with overlapping lazy quantifiers, leading to inefficient regex complexity. The vulnerability allows an authenticated AMQP 1.0 consumer with read and write permissions on stream queues to cause increased CPU usage by submitting crafted LIKE filters combined with messages having long property values. The regex engine's default match limit caps each match to approximately 100-200 milliseconds, but this can be multiplied across many messages and sessions, resulting in substantial CPU amplification. The vulnerability is addressed in rabbitmq-server versions 4.2.7 and 4.3.1.

An attacker with authenticated AMQP 1.0 consumer access and read+write permissions on stream queues can cause elevated CPU usage on the server by exploiting inefficient regex evaluation of LIKE filters. This can degrade performance by consuming CPU resources disproportionately per message delivered, especially under high message throughput and multiple parallel sessions. There is no indication of data compromise or denial of service beyond resource consumption.

This vulnerability is fixed in rabbitmq-server versions 4.2.7 and 4.3.1. Users should upgrade to these or later versions to remediate the issue. No additional mitigation is required if these versions are in use.