Server and hypervisor monitoring
Server monitoring runs around the clock on all plans and does not use engineer hours. The plan defines how deep the checks go.
How it works
- The Zabbix agent in active mode sends metrics to the monitoring server in the Netherlands.
- Monitoring runs 24/7.
- Alerts from the monitoring system go to INTROSERV engineers, who analyze them. If an alert requires action to fix its cause, we create a ticket for you with a description of the problem and options for solving it.
Monitoring covers network availability via ICMP, OS and hardware parameters (RAID, SMART, temperature), and services and applications. The depth and scope of monitoring depend on your plan.
What is monitored on each plan
| Plan | Checks | Why |
|---|---|---|
| Starter | network availability via ICMP: response, response time, packet loss | Confirms that the server is running and reachable from the network. |
| Advanced | everything in Starter, plus CPU and memory load, swap | Lets us notice a lack of server resources before the server stops responding. |
| disk usage and disk I/O | Warns about low disk space and slower disks. | |
| network traffic | Shows the load on the network link. | |
| hardware temperature sensors | Warns about hardware overheating. | |
| software and hardware RAID status, including ZFS on Linux and Proxmox VE | Reports a disk array fault, for example when one of the disks fails. | |
| proactive notifications | You learn about a problem before it causes a failure. | |
| Premium | everything in Advanced, plus hardware RAID controller details: firmware version, battery and cache status | Helps detect RAID controller battery wear and faults in advance. |
| service status | Reports when services important to you stop. | |
| custom triggers and thresholds on request | Monitoring is tuned to the specifics of your server and applications. |
Services monitored on Premium
We check the state of each service: whether it is active or has terminated abnormally.
| System | System services | Database services |
|---|---|---|
| Linux and Proxmox VE | cron, dbus, rsyslog, ssh, systemd-journald, systemd-networkd, systemd-resolved, systemd-udevd, systemd-timesyncd, unattended-upgrades | postgresql, mariadb, mysql, mongod, redis-server, valkey-server |
| Windows Server | EventLog, RpcSs, LanmanServer / LanmanWorkstation, W32Time, Netlogon, Schedule, WinRM, TermService, MpsSvc / BFE, WinDefend | PostgreSQL, MariaDB, MySQL 8.x, MongoDB, Redis |
Let us know about other services that are important to you during onboarding. On Premium, they can be added as custom triggers.
Typical alerts
- disk space is running low;
- a service has stopped;
- load is growing;
- a RAID array is degraded;
- overheating.
On all plans, we fix the causes of monitoring alerts with your approval.
Info
Engineers solve hardware and network problems on their own, without approval. Work inside the OS is done only through a ticket and after your approval.
Proxmox VE hypervisor monitoring
On Proxmox VE, hypervisor monitoring works in addition to server monitoring. Each node is enabled separately and billed under its own plan.
Monitoring covers local, network and distributed storage, VM status, backup jobs and cluster status.
| Plan | What is checked | Why |
|---|---|---|
| Starter | local storage | Warns about low space for virtual machines. |
| VM status: running, stopped, paused, restarting | Reports an unplanned stop or restart of a virtual machine. | |
| VM configuration changes | Records unplanned changes to virtual machine settings. | |
| QEMU guest agent availability | Reports that the hypervisor has lost contact with the agent inside a virtual machine. | |
| Advanced | everything in Starter, plus network storage | Reports a lost connection to network storage. |
| CPU, memory and storage overcommit | Warns when virtual machines are allocated more resources than the server has. | |
| backup job status, overall and per VM | Shows which virtual machines have no recent backup. | |
| vmbr interface status | Reports a network connection failure for virtual machines. | |
| PBS datastore status | Reports problems with the Proxmox Backup Server datastore. | |
| Premium | everything in Advanced, plus Ceph status: OSD, PG, RBD latency | Warns about reduced reliability and performance of distributed storage. |
| HA status: quorum, fencing, pve-cluster | Warns about the risk of losing cluster fault tolerance. | |
| ZFS replication | Reports a failure or lag in data replication between nodes. | |
| Corosync links and latency | Reports broken communication between cluster nodes. |
Info
Engineers fix hardware and network incidents on their own. Work at the hypervisor and VM level is agreed with you.
Triggers and thresholds
An alert fires when a parameter goes beyond its threshold and stays there for a set time. Some parameters have two levels: warning and critical. The default values are listed below.
Server: Linux, Windows Server and Proxmox VE nodes
Proxmox VE nodes are checked against the same triggers as Linux servers, plus the hypervisor triggers in the next table.
| What is monitored | Default threshold |
|---|---|
| Network availability | no response to 3 checks in a row; packet loss over 20% or response time over 150 ms for 5 minutes |
| Agent and reboot | no data from the agent for more than 5 minutes; server uptime less than 10 minutes |
| CPU | load over 90% for 15 minutes; on Linux, load average over 1.5 per core, on Windows, processor queue over 3 per core for 5 minutes |
| Memory | over 90% used for 5 minutes; on Linux, less than 50% of swap free, on Windows, page file over 80% full |
| Disk space | over 80% full is a warning, over 90% is critical; on Linux, also free inodes below 20% and 10% |
| Disk performance | disk utilization over 95% or response time over 30 ms for 15 minutes |
| Network | an interface has lost its link; more than 2 errors per second for 5 minutes |
| Temperature | over 80 °C is a warning, over 95 °C is critical |
| Disks and arrays | a disk, RAID array or ZFS pool has failed; a ZFS pool is over 80% full |
| RAID controller | the controller or cache battery has failed; the cache has switched to WriteThrough mode; temperature over 95 °C; the firmware version has changed |
| Services | a service has stopped; a service has been restarted |
Proxmox VE hypervisor
| What is monitored | Default threshold |
|---|---|
| Virtual machines and containers | a VM is not running or has been restarted; the QEMU guest agent is not responding; the VM state or configuration has changed |
| Node resources | CPU or memory over 90% for 15 minutes; VMs are allocated more than 300% of the node's cores or more than 100% of its memory |
| Storage | storage is unavailable; usage over 80% and 90%; VM disks are allocated more than 150% of the storage capacity |
| VM network | a vmbr bridge is not active |
| Backup | a backup job finished with an error or warning; no recent VM backup for more than 26 hours; backup verification failed; a VM is not included in any backup job |
| Cluster and HA | a node has lost quorum; not all nodes are online; no HA quorum; fencing is in progress; HA services are in an error state; a Proxmox VE system service has stopped |
| Ceph | HEALTH_WARN or HEALTH_ERR status; an OSD is down; some data is unavailable or not recovered for more than 15 minutes; usage over 75%; OSD latency over 50 ms |
| ZFS replication | replication finished with an error; the last successful sync is older than 1 hour |
| Corosync | a link between nodes is down or unstable; latency over 5 ms |
Info
The values in these tables are default thresholds. On Premium, thresholds and custom triggers are tuned to your server on request, including for specific disks, sensors, interfaces and storage.
