Hey all,
Relatively new installation of 3 node cluster, running CEPH for storage. over the course of about two weeks one of the nodes went from ~40GB of ram used to 265. Moved the VM's around so that we could reboot the node, and it went back to the "normal" usage (with the machines on) of about 28. Over the next week with no new VM's running, we're up to 68.
Attached are outputs of
top -o %MEM -b -n 1
cat /proc/meminfo
cat /proc/spl/kstat/zfs/arcstats
Relatively new installation of 3 node cluster, running CEPH for storage. over the course of about two weeks one of the nodes went from ~40GB of ram used to 265. Moved the VM's around so that we could reboot the node, and it went back to the "normal" usage (with the machines on) of about 28. Over the next week with no new VM's running, we're up to 68.
Attached are outputs of
top -o %MEM -b -n 1
cat /proc/meminfo
cat /proc/spl/kstat/zfs/arcstats