Hi,
I have been monitoring our memcached servers with packetbeat. This was noticed after deploying packetbeat 5.4.0, it may have been happening before, but I'm not 100% sure. I will consistently get packetbeat taking all of the cpu on the server. This impacts the performance of memcache. For now, we have shutoff packetbeat.
I have 4 instances of memcache running on the server. Top looks like this before packetbeat.
top - 07:51:32 up 100 days, 6:35, 1 user, load average: 0.19, 0.13, 0.05
Tasks: 164 total, 1 running, 163 sleeping, 0 stopped, 0 zombie
Cpu(s): 0.9%us, 3.5%sy, 0.0%ni, 94.3%id, 0.1%wa, 0.0%hi, 1.1%si, 0.0%st
Mem: 8054356k total, 7672104k used, 382252k free, 65656k buffers
Swap: 8388604k total, 20220k used, 8368384k free, 4616300k cached
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
20020 memcache 20 0 1069m 778m 712 S 6.3 9.9 1832:06 memcached
20052 memcache 20 0 1067m 783m 712 S 6.0 10.0 1839:16 memcached
6470 memcache 20 0 558m 190m 708 S 3.3 2.4 2187:18 memcached
13 root 20 0 0 0 0 S 0.3 0.0 29:47.84 ksoftirqd/2
1419 root 20 0 63608 3716 3300 S 0.3 0.0 130:39.73 vmtoolsd
6964 root 20 0 366m 8524 5308 S 0.3 0.1 11:39.54 pxp-agent
Starting up packetbeat it now looks like this.
top - 07:52:45 up 100 days, 6:36, 1 user, load average: 0.32, 0.19, 0.08
Tasks: 166 total, 1 running, 165 sleeping, 0 stopped, 0 zombie
Cpu(s): 24.3%us, 4.9%sy, 0.0%ni, 68.7%id, 0.1%wa, 0.0%hi, 2.0%si, 0.0%st
Mem: 8054356k total, 7778452k used, 275904k free, 65912k buffers
Swap: 8388604k total, 20220k used, 8368384k free, 4616040k cached
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
20913 root 20 0 457m 110m 10m S 77.1 1.4 0:37.20 packetbeat
20020 memcache 20 0 1069m 778m 712 S 9.6 9.9 1832:12 memcached
20052 memcache 20 0 1067m 783m 712 S 7.3 10.0 1839:21 memcached
6470 memcache 20 0 558m 190m 708 S 1.0 2.4 2187:20 memcached
52568 root 20 0 4374m 144m 15m S 0.3 1.8 169:59.43 java
1 root 20 0 30424 1704 1424 S 0.0 0.0 0:09.85 init
The packetbeat.yml is configured like this.
packetbeat.interfaces.device: any
packetbeat.flows:
timeout: 30s
period: 30s
packetbeat.protocols.memcache:
ports: [11215, 11216, 11217, 11223]
output.elasticsearch:
hosts: ['ingest-logcentral.dx.deere.com:9200']