On Thu, Oct 15, 2015 at 6:04 PM, Alaa Zubaidi (PDF) <alaa.zuba...@pdf.com> wrote:
> We are running Cassandra 2.2.1 on Windows 2008R2, and we see that multiple > Nodes are stuck at 99% CPU bringing the whole VM to a halt. > We suspect that there is another process that IT/Windows is causing the > CPU issue, but the problem is Cassandra does NOT recover, the CPU > utilization start climbing until the VM is not usable. If we restart > Cassandra, things go back to normal. > Most cases where a JVM does not recover and churns at maxed CPU are the result of GC failure and/or OOM. Check your logs for OOM and long GCs. Also FWIW you are among a relatively small group of Windows operators. Other than with the people working at datastax to support Windows, there is not a whole lot of well understood operational best practice for Cassandra on Windows. =Rob
