Closed rps-v closed 7 years ago
hi @rps-v, we will investigate this.
@rps-v it seems that this is a bug in Sigar, a plugin that we use for cpu and memory monitoring. Can you tell us when this crash happens (on heavy load, etc.) ?
A workaround would be if you remove the sigar
directory from the plugins located in $CRATE_HOME/plugins/
. We encountered a bug that prevents you from starting a node after removing the folder (see https://github.com/crate/crate/commit/408c6a117d555b3782dfd2db6891f5e8fbe49efb). In your case you need to remove the sigar-path as shown in the commit manually in your crate.bat
.
@mikethebeer It was a one time crash and it happened overnight (should have been very light / no load on the server). The setup was a machine running 2 instances of Crate (a cluster with 2 nodes).
Here is a snippet from the logs at the moment of the crash (doesn't contain much): crate.log.txt
Thanks for the quick response and workaround, will report back if it happens again !
@rps-v thanks, tell us if it happens again 👍 probably it has something to do with https://github.com/hyperic/sigar/issues/77
CrateDB version: 1.0.4
JVM version: 25.121-b13 (JRE: 8.0_121-b13)
OS version / environment description: Windows Server 2012 R2
Problem description: Fatal crash
Log: hs_err_pid9872.log.txt
NOTE: There is an additional .mdmp file (~700mb) if needed
Steps to reproduce:
Unfortunately I do not have any logs of what was executing (SQL-wise), just the crash log.