Check system statistics

There are some system statistics that might help with troubleshooting.

Host resource usage

To check the CPU, memory, and disk usage on a Linux machine, you can run the following command:

sudo echo -n "CPUs: ";cat /proc/cpuinfo | grep processor | wc -l; echo -n "Mem: ";free -h | grep Mem | tr -s " " | cut -d" " -f 2; echo -n "Disk: "; df -h / | grep -v Filesystem;
CPUs: 72
Mem: 251G
Disk: /dev/sda2       160G   13G  148G   8% /
10.1.12.104
CPUs: 88
Mem: 251G
Disk: /dev/sda2       208G   22G  187G  11% /
10.1.12.105
CPUs: 88
Mem: 251G
Disk: /dev/sda2       208G  5.1G  203G   3% /

Generally, common tools such as top or iostat may be useful.

Auditd

If top reports high CPU usage for the auditd process, it may have rules auditing some system calls frequently used by YugabyteDB which can significantly affect performance. You can try temporarily disabling audit by running the following command on each YugabyteDB node:

auditctl -e 0

Then you would check if this improves performance.

To re-enable audit afterwards, run the following command:

auditctl -e 1

YugabyteDB processes state

YugabyteDB provides the following web endpoints where the current state of each process is aggregated. This includes logs, flags, as well as memory, disk, and network usage metrics. Additionally, it provides dedicated metrics endpoints for YCQL and YSQL requests:

Description URL
Master Web Page <node-ip>:7000
TServer Web Page <node-ip>:9000
YCQL Metrics <node-ip>:12000/metrics
YSQL Metrics <node-ip>:13000/metrics

When running yb-ctl locally with default values, three local IP addresses are created: 127.0.0.1, 127.0.0.2, and 127.0.0.3, one for each YugabyteDB node.