Home
Where Keal opens: how your account stands as a whole. What concerns one stack or one VM is in its own pane, on the Stacks and VMs tabs.
Your fleet
Keal runs VMs on a fleet of physical servers - its nodes - which boot from the engine over the network. An administrator's Home opens on how hard that fleet is working: four gauges for the fleet as a whole, and a line for each node with the same figures for it. A gauge is green up to 70%, amber up to 90% and red beyond. Only nodes that are running count towards what the fleet has; the others are listed, and say what they are doing.
CPU pressure
How busy the processors are: the nodes' load - the number of processes running or waiting to run, averaged over a minute - against the cores they have. 100% is every core busy; more than that is work waiting for a core.
Memory
How much of the nodes' memory is in use - by the VMs on them, and by the nodes themselves.
Storage
How much of the fleet's disk space is in use: the nodes' own storage and the engine's storage pools, each disk counted once.
vCPUs given out
How much of the room for vCPUs is taken. VMs seldom use all their vCPUs at once,
so a node may be given more vCPUs than it has cores: as many per core as the
engine's CPU_OVERCOMMISION setting says (Engine settings), which is also how
Keal decides which node a VM can start on - a node takes one while its VMs'
vCPUs, and the new one's, fit. Counted here are the vCPUs of the VMs running on
the nodes, and those held for VMs that are shut off but keep their disks on a
node, so can only start there. At 100% there is no room for another vCPU.
Free for new VMs
The biggest VM there is room for on any one node, by the rules Keal places VMs by: its memory must leave 512 MB free on the node, and its vCPUs must fit in the node's room for them (see vCPUs given out) once the vCPUs held for its stopped VMs are counted. A node that is asleep counts - it is woken for the VM - and one in maintenance does not.
Below it, the biggest that fits with GPUs: on a node with GPUs free that its VMs have not been given, which may be another node than the first. A fleet with no GPUs says so - a VM can only be given a GPU on a node that has one.
Busiest VMs
The VMs working hardest right now, five of each, with a line for the last hour, a point every five minutes: by CPU - how much of their own vCPUs they used over the last ten minutes - and by network traffic, in and out over the last hour. An administrator sees every VM; anyone else, their own.
Status
The first thing Home shows: a card for each kind of thing you have, and one that says whether it is all well. The figures are counted from the same lists as the tabs themselves, and a card takes you to its tab.
| Card | What it counts |
|---|---|
| Stacks | Your stacks and how many are running, and any VMs that are not part of a stack. |
| VMs | Your VMs, how many are running, and the vCPUs and memory the running ones hold. |
| Nodes | The engine's nodes, their cores and how much memory is free. Administrators only. |
| Storage | The disk space your images take up now, and the size they could grow to. |
| Networks | Your connections, and how many carry an external address. |
Health gives one word for each of five things. Click a line to see what the word was made of, with links to whatever needs looking at.
- VMs - Healthy unless one has crashed. A VM you shut down or paused is not a fault.
- Monitoring - whether the checks on your VMs pass. A failing check you have acknowledged shows as Known issues rather than as failing. See Monitors.
- Updates - what your VMs last reported was waiting to be installed; security fixes and a pending restart are called out. See Software.
- Backups - whether the images your VMs run from have been backed up in the last 24 hours. An image less than a day old has not missed anything yet. If any are missing you are also told so when Keal opens, whichever tab it opens on. See Images.
- Nodes - whether every node answers and has room left on its storage. A node put to sleep, shut down or taken into maintenance is not a fault. Administrators only.
The cards refresh by themselves while you are looking at them.
The same figures are in a terminal as keal status - see the command line.
Usage
What your account has in use, against the quotas it is allowed - vCPUs, memory, storage, external addresses and network traffic. A bar turns amber at three quarters of a quota and red at nine tenths; a quota shown as ∞ has no limit.
Now is what is in use at this moment. The other choices are a past month's average, which is what a month is billed on. Where the engine shows costs, each line has its cost per month and the table its total.
CSV downloads the same figures for each month of the year so far, up to the month picked, for a spreadsheet.
Backup storage is an estimate, and network traffic is only counted for the account as a whole.
Software
What your VMs run: each one's operating system and hostname, and the software installed on them, counted across VMs - which is how to find every VM still on an old version of something.
The operating systems are known for every VM. The installed software is only known for VMs that have been scanned, which is done a stack or a VM at a time from its own pane, on the Stacks or VMs tab. Where a scan found updates, Install updates and Install security updates install them on every running VM that can be asked, each VM doing its own installing. Clear here forgets what every scan found. CSV downloads either list.
Monitors
Every check on every VM you have, with what needs looking at first: what is down, then what is down and known about, then the rest. Monitors are added from a VM's or a stack's own pane; pausing, acknowledging and removing can be done from here. See a VM's monitors for what the checks are.
Uptime
For a month, how much of the time each monitor was being checked it was up - all your VMs together, with a CSV of the month shown.