Lei,
I have seen this on 1.9.2 standalone and cluster deployments. Every time I have seen this so far, the reason has been an input port in a process group which is not connected by an output port. The input port can be identified in the user interface by drilling down to the active thread. Disabling or removing the input port resolves the high CPU usage. Warm regards, <https://www.alertlogic.com/> Jim Williams | Principal Database Developer O: +1 713.341.7812 | C: +1 919.523.8767 | jwilli...@alertlogic.com | <http://www.alertlogic.com/> alertlogic.com <https://twitter.com/alertlogic> <https://www.linkedin.com/company/alert-logic> From: wangl...@geekplus.com.cn <wangl...@geekplus.com.cn> Sent: Monday, September 16, 2019 4:53 AM To: users <users@nifi.apache.org> Cc: dev <d...@nifi.apache.org> Subject: NiFi cluster heavy cpu usage I deployed a nifi cluster with two virtual machines, each 4 core, 16 GB memroy The cpu is more than 100% even there's no processor running. Often the cpu is about 300% after i start some processors. [root@iZuf6agrzcolqeja3if7kbZ ~]# ps -mp 3429 -o THREAD,tid,time | sort -k2 -n -r |less root 171 - - - - - - 07:08:43 root 16.1 19 - futex_ - - 3598 00:40:18 root 16.1 19 - futex_ - - 3569 00:40:29 root 16.0 19 - futex_ - - 3597 00:40:02 root 15.9 19 - futex_ - - 3603 00:39:51 root 15.9 19 - futex_ - - 3601 00:39:57 root 15.9 19 - - - - 3604 00:39:57 root 15.8 19 - - - - 3713 00:39:40 root 15.7 19 - - - - 3600 00:39:27 root 15.6 19 - futex_ - - 3712 00:39:00 root 15.6 19 - futex_ - - 3593 00:39:05 There's some threads consumes cpu. I pick one and jstack it, often it is in waiting state. Any insight on this? Thanks, Lei _____ wangl...@geekplus.com.cn <mailto:wangl...@geekplus.com.cn>
smime.p7s
Description: S/MIME cryptographic signature