What Monitor reads
Open Monitor › Overview to see what Monitor reads and change it. Changes go through one review. You need to be an Admin.
What it collects
Section titled “What it collects”| Collector | Reads |
|---|---|
| Health, cycles and memory | Status, balance, runway and memory. Always on. |
| Where cycles go | Spending by compute, storage, calls and more. |
| Snapshots | Count and age of each canister’s snapshots. |
| Change history | Code and controller changes. |
| Logs | Errors and traps from canisters that share their logs with the Agent. |
| Subnet health | Node performance of the subnets your canisters run on, hourly. |
How often
Section titled “How often”How often sets how often each canister is read: every 5, 15 or 30 minutes, or every hour. Agent setup starts at every 5 minutes. Reading more often costs more of the Agent’s cycles.
Under Advanced, choose Which canisters: all of them, or only some. Log settings are there too: Read logs every and a Daily sampling allowance per canister.
What the Agent can read
Section titled “What the Agent can read”What Monitor can collect depends on the access each canister gives the Agent. Overview groups your canisters:
- Full access: the Agent is a controller, so it reads everything.
- Status only: the Agent can read status, but not spending or snapshots.
- Public data only: the Agent has no access yet. Give it access from Settings and access.
Check it’s working
Section titled “Check it’s working”After you start Monitor, wait for the first round, then look at Overview: it shows when each round finished and any canister it couldn’t read. Choose Read all now to read straight away.
Good to know
Section titled “Good to know”- A canister that can’t be read gets a gap, never a fake zero.
- Large fleets are read in batches, so a round can take longer than the interval. Overview shows any canisters that are overdue.
- For a canister that could run out of cycles quickly, read it every 5 minutes and keep enough balance to react.