yep, so far it looks like a file descriptor leak. Not sure if gc or
some other event like compaction would close these files..
[root@CAP-VM-1 ~]# ls -al /proc/31134/fd | grep MSA | wc -l
540
[root@CAP-VM-1 ~]# ls -al /proc/31134/fd | grep MSA | wc -l
542
[root@CAP-VM-1 ~]# ls -al /proc/31134/
I'm seeing this also, and my nodes have started crashing with "too many open
file errors". Running lsof I see lots of these open tmp files.
java 8185 root 911u REG 8,32 38
129108266
/opt/cassandra/data/MonitoringData/Properties-tmp-hc-268721-CompressionI
