In most cases, minor OS upgrades are not problematic provided you have sufficient capacity in your cluster so that it can tolerate scheduled downtime while some nodes are being upgraded.
One thing you should be aware of are patches in newer versions of Linux distributions which address Spectre and Meltdown vulnerabilities. The microcode and security patches in RHEL 7.7 can have a significant impact on the performance of servers. Note that this is not limited to Cassandra workloads -- the fixes are applied at the OSI layer so it affects all applications and databases running on patched operating systems. You can find out more information about the RHEL patches in this KB -- https://access.redhat.com/articles/3311301. Some recommended mitigation strategies for the performance impacts include: - Reviewing disk configuration, particularly setting a low(er) read-ahead size. See this blog post from The Last Pickle for guidance -- https://thelastpickle.com/blog/2019/01/30/new-cluster-recommendations.html . - Disabling governors which use a lower CPU frequency, or explicitly setting CPU scaling to "performance". For reference, I previously wrote about this here -- https://support.datastax.com/hc/en-us/articles/115003018063. - Set the disk access mode to mmap index files only. I've explained the details in this post -- https://community.datastax.com/questions/6947/. Cheers!
