Portworx integration with prometheus

PX storage and network stats can easily be integrated with prometheus or similar applications. These stats are exported at port 9001; your application can poll http://<IP_ADDRESS>:9001/metrics to get their runtime values.

Integration with Prometheus

Step 1: Configuring Prometheus to watch px node

Add your px node as a target in Prometheus config file:

Prometheus Config File

In the example above, our node has IP address of 54.173.138.1, so Prometheus is watching 54.173.138.1:9001 as its target. This can be any node in the PX cluster.

Step 2: PX metrics to watch and building graphs with Prometheus

Once Prometheus starts watching px node, you will be able to see new portworx related metrics added to Prometheus.

PX Metrics in Prometheus

You can now build graphs:

Building a Graph with Prometheus

Note

A curl request on port 9001 also shows the stats:

Curl Request on 9001

Storage and Network stats

Cluster stats:

px_cluster_cpu_percent: average CPU usage for the PX cluster nodes in percentage
px_cluster_disk_available_bytes: available storage in px cluster in bytes
px_cluster_disk_utilized_bytes: used storage in px cluster in bytes
px_cluster_memory_utilized_percent: average memory usage for the px cluster nodes
px_cluster_pendingio: total bytes (read/write) being currently processed
px_cluster_status_cluster_size: total cluster size
px_cluster_status_nodes_offline: total offline nodes
px_cluster_status_nodes_online: total online nodes
px_cluster_status_nodes_storage_down: total storage down nodes
px_cluster_status_cluster_quorum: cluster_quorum, 1 = in quorum, 0 = not in quorum
px_cluster_disk_total_bytes: total storage in px cluster in bytes
px_cluster_status_storage_nodes_online: number of nodes proving storage which are online (these participate in quorum)
px_cluster_status_storage_nodes_offline: number of nodes proving storage which are offline (these participate in quorum)

Node stats

px_network_io_bytessent: bytes sent by this node to other nodes
px_network_io_received_bytes: bytes received by this node from other nodes
px_node_status_<node_id>_status: <node_id> status, 1 = online, 0 = offline
px_node_stats_cpu_percent_usage: last reported CPU usage
px_node_stats_free_mem: amount of inactive and idle memory (reported by /proc/vmstat)
px_node_stats_total_mem: the amount of idle memory (reported by /proc/vmstat)
px_node_stats_used_mem: total - free

Volume stats

px_volume_capacity_bytes: volume size
px_volume_depth_io: number of i/o operations being served at once
px_volume_halevel: volume HA level
px_volume_iops: operations per second
px_volume_read_bytes: total bytes read from volume
px_volume_reads: number of read operations served by the volume
px_volume_readthroughput: bytes read per second
px_volume_writes: number of write operations served by the volume
px_volume_writethroughput: bytes written per second
px_volume_written_bytes: total bytes written to the volume
px_volume_dev_depth_io: I/Os currently in progress as reported by block device
px_volume_dev_read_latency_secs: read latency for block device (total time spent reading/ number of reads)
px_volume_dev_readthroughput: read throughput for block device
px_volume_dev_write_latency_secs: write latency for block device (total time spent writing/ number of writes)
px_volume_dev_write_throughput: write througput for block device
px_volume_iopriority: configured volume io_priority (0 = low, 1 = medium, 2 = high)
px_volume_fs_capacity_bytes_bytes: total size reported by filesystem
px_volume_fs_usage_bytes_bytes: used capacity reported by filesystem

Disk stats: collected from /proc/diskstats

px_disk_stats_interval_seconds: interval seconds
px_disk_stats_io_seconds: time spent doing I/Os
px_disk_stats_progress_io: I/Os currently in progress
px_disk_stats_read_bytes: read bytes
px_disk_stats_read_seconds: time spent reading
px_disk_stats_reads: reads
px_disk_stats_used_bytes: used bytes
px_disk_stats_write_bytes: write bytes
px_disk_stats_write_seconds: write seconds
px_disk_stats_writes: writes
px_disk_stats_read_latency_seconds: read latency for disk  (read ms / number of reads)
px_disk_stats_write_latency_seconds: write latency for disk (write ms / number of writes)

Portworx Alerts