We are using one of the more recent versions of Metricbeat - v8.1.1 for 2 types of instances. Cloud Instances (Ubuntu 22.04) and Bare Metal installations (Debian 11).
I have observed specifically that the metrics under - system.process.cgroup.io.pressure.* are not reported for the Bare Metal Debian 11 instances, but come in correctly for the Ubuntu (Cloud-based instances).
The RAM and CPU Cgroup Pressure Metrics are available though - for both the aforementioned deployments.
The metricbeat - modules.d/system.yml file looks the same for both.
- module: system
period: 30s
metricsets:
- cpu
- load
- memory
- network
- process
- process_summary
- socket_summary
- network_summary
- core
- diskio
- socket
process.include_top_n:
by_cpu: 5 # include top 5 processes by CPU
by_memory: 5 # include top 5 processes by memory
- module: system
period: 5m
metricsets:
- filesystem
- fsstat
filesystem.ignore_types: [ssysfs, tmpfs, bdev, proc, cgroup, cgroup2, cpuset, devtmpfs, debugfs, tracefs, securityfs, sockfs, bpf, pipefs, ramfs, hugetlbfs, devpts, mqueue, pstore, autofs, efivarfs, configfs, fuse, fusectl]
processors:
- drop_event.when.regexp:
system.filesystem.mount_point: '^/(sys|cgroup|proc|dev|etc|host|lib|snap)($|/)'
- module: system
period: 1m
metricsets:
- service
service.pattern_filter: ["ssh.service"]
- module: system
period: 15m
metricsets:
- uptime
I have started Metricbeat in the foreground using metricbeat -e -d '*' as well as suggested in some of the forums, but even there these metrics seem to be missing.
The following image is what I see when I query for system.process.cgroup.io for all the fields
I have also checked that the Debian 11 systems have Cgroups V2 available and the Kernel version there is - 5.10.0-8-amd64
Can I understand what could be the issue here?
The Cgroup process level I/O pressure information is important for understanding systemd level resource utilization and alerting for me.