gromitor is real-time, cross-platform container resource monitoring as a lightweight saas. drop one agent on docker or kubernetes and watch live cpu, memory, network, and disk per container — no heapster, prometheus, or grafana to deploy and babysit.
as one engineer put it on hacker news: “there are open source tools like heapster which have to be manually deployed.” that's the whole pain. before you can see a single number you're standing up a metrics pipeline, running storage, wiring dashboards, and keeping it all patched — across every host and cluster.
gromitor is the hosted layer so you don't run the tools. install one lightweight, read-only agent and we run ingestion, storage, the real-time dashboard, and alerting. your containers show up in seconds — docker and kubernetes, side by side, in one pane.
copy one curl line for docker or one kubectl apply for kubernetes. the gromitor agent comes up read-only, registers itself, and starts streaming — no dashboards to deploy, no storage to run.
within seconds your containers populate the dashboard with live cpu, memory, network i/o, and disk i/o. recent history fills in as sparklines so you see the shape, not just a number that resets.
create a simple rule — cpu over 80% for 5 minutes — and gromitor watches it for you. when it breaches you get an in-app and email alert, and again when it resolves.
every container reports cpu %, memory, network i/o, and disk i/o on a tight loop. the dashboard streams it in with sparklines for the last 15–30 minutes, so you read the trend at a glance.
click any container for a granular detail view — the recent shape of each metric, the image it's running, and where it lives.
the same lightweight agent speaks both engines. monitor a single docker host and a multi-node kubernetes cluster side by side, in one list, with one mental model.
no heapster to stand up, no prometheus to operate, no grafana to wire — gromitor is the hosted layer so you don't run the tools.
set plain rules — a metric, an operator, a value, a duration — per container or across an agent. gromitor evaluates them continuously against the live stream.
breaches arrive in-app and by email, with a matching resolved notice when things calm down, so a busy hour doesn't bury the one alert that mattered.
gromitor reads cpu, memory, and i/o straight from each engine's own stats api — read-only, batched over https, under ~2% host cpu. it never reads what's inside your containers, only their resource counters. recent history lives for ~24–48h, which is exactly what live monitoring and short-window alerts need.
one agent, one pane, zero open-source tools to deploy. free to get started.