Monitoring

PDM constantly watches whether stands are working and keeps the history of their availability. Thanks to that, “is the stand down or is it just me?” can be answered without going into the cluster or asking colleagues.

What exactly is tracked

PDM polls the clusters regularly and, for each stand, counts how many of its applications are actually running compared to how many should be. The status of the stand comes out of that ratio.

Status	When it happens
Operational	Everything expected is running
Degraded	Some of the applications are running
Down	No application is running
Unknown	The cluster did not answer the check — there is no data

“Unknown” and “Down” are different things

A grey status means PDM could not get an answer from the cluster: the network, access rights, an overloaded cluster. The stand itself may be working perfectly at that moment. That is why “no data” gaps are not counted as downtime in the availability statistics but are shown separately.

Monitoring answers the question “are the stand's applications running”, not “are they working correctly”. An application can be running and still answer with errors — PDM will show such a stand as green.

Where you can see it

The coloured dots in the sidebar give an overview of every environment at once; in the stand header Uptime is shown next to the status — the share of time the stand was available over a period (30 days by default).

The most detailed picture is on the Dashboard tab: the current state with the time of the check, uptime over the period, service readiness, Availability history with a 7 days / 30 days period switch, and Recent status changes — up to ten most recent transitions with date and time.

How to read the indicators

• Uptime is the share of time within the selected window when the stand was available. If there was no data for a significant part of the window, a “No data: N.N%” line appears under the percentage — take it into account, otherwise the percentage can be read far too optimistically.
• Service readiness explains the “Degraded” status: “7/9” means that seven out of nine expected applications are running and two are not. It is worth checking against the Services tab to see which application was changed recently.
• Recent status changes is not a list of incidents with causes, it is a record of transitions. It is a convenient way to answer “when did it break” and match the time against a recent deploy.

How often the data is refreshed

A check runs roughly once a minute and the result reaches open pages immediately — there is no need to refresh them. The availability chart is filled in less frequently: the data accumulates and is written in batches.

The availability history and the list of status changes are kept for a limited time; the retention depth is set by the administrator, so there may be no data for a very old period.

If a stand is shown as down

First work out what exactly you are looking at.

1. The “Unknown” status. There is no data — the problem is most likely with the connection to the cluster rather than with the stand. If the time of the last check is old, monitoring cannot reach that cluster right now. Try opening the stand's address from the Hosts panel — that is a direct check of whether the environment works.
2. The “Degraded” status. Some of the applications did not come up. Look at the Service readiness card to gauge the scale, and at the status change feed to see when it started.
3. The “Down” status. Nothing is running. That happens right after a deploy, while the applications are still starting, and also when a stand has been stopped deliberately.

Next it is worth answering the question “what changed”.

• Open the Workspaces section and check whether a configuration has been applied to this stand recently and how the operation ended: a failed step explains why an application did not come up.
• On the Services tab, check for fresh overrides — a version of an application that has just been set, for example.
• Match the time from the Recent status changes feed against the time of the deploy.

If the stand does not come up on its own and the cause is not the last configuration change, the question moves to the infrastructure side — contact the team responsible for that cluster and tell them specifically: the stand name, the cluster, the time and the status PDM is showing.

Do not rush to change the configuration “just in case”

While the cause is unclear, applying settings again only adds variables: it will overwrite values and restart the applications. Look at the status history and the recent workspaces first.

What's next

• Stand page — the stand's tabs and the addresses of its services.
• Stand configuration — how to see what was changed on the stand.
• Workspaces — how to work through the steps of a failed operation.