Pivotal Cloud Foundry® (PCF) Elastic Runtime
If you see issues where VM's are using unusually high amounts of memory, you may be affected by the following bug where Rsyslogd uses excessive amounts of memory.
There is a memory leak in Rsyslogd 8.23.0 which is the version used in PCF1.8.21-build.1 in Stemcell 3263.12.
For example, check for errors like the following in your message logs (/var/log/messages):
Jan 3 16:39:24 localhost vmsvc: [ warning] [guestinfo] Failed to get vmstats. Jan 3 16:39:24 localhost rsyslogd: action 'action 0' resumed (module 'builtin:omfwd') [v8.23.0 try http://www.rsyslog.com/e/2359 ]
You can check the version of rsyslogd that you are using by SSH'ing onto a VM and running the following command: rsyslogd -version
You will need to upgrade your Stemcell to version 3263.15 or later. In this version, the Ubuntu stemcells have been PIN'ed to rsyslog to 8.22.0.
To check the amount of memory being consumed by rsyslogd, please run the following command:
ps auxw | grep rsyslogd | grep -v grep