It would be useful to have a "Test Alert" button for each configured alert in order to integrate and test alerts with third-party systems. Otherwise, it is difficult if not impossible to determine what the alert will look like until it is triggered.

Collect hardware metrics even if there's no managed mongo process

Have Automation Agent collect hardware metrics on unmanaged mongo hosts.

Automation agents doesn't collect hardware metrics unless there's a managed mongo process. This means we can't provide centralized system monitoring for a heterogeneous environment, where some clusters are running on their own and others are under automation, or on any non-managed host.

please provide an option in Ops Manager to monitor the balancer status and add to alerts?
So, we will know if balancer is not running.

Note:
Normally the balancer would be disabled during backups and during a scheduled Balancing Window downtime.
I believe that the Balancer has a duty cycle of either 10 secs when nothing recently to balance, or 1 sec when there's a bunch of balancing to do.
Any alert would need to account for these:
the changelog shows chunk moves commanded
the actionlog shows balancer state change history
the settings collection has the balancer state

What is the problem that needs to be solved? Allow to configure maxTimeMS for MongoDB commands which are executed from Ops Manager's Data Explorer.

Why is it a problem? (the pain) A) Ops Manager's Data Explorer cannot work with [views](https://docs.mongodb.com/manual/core/views/index.html) in case if the view is taking >15000 ms to be load. Data Explorer cannot work with [find](https://docs.mongodb.com/manual/reference/command/find/index.html) operations in case if that find operation is taking >15000 ms to be completed.

Ops Manager users with hundreds or even thousands of replica set members (hosts/nodes) need an alert that indicates a nodes is in RECOVERING state and is too far behind the oplog to recover without manual intervention. This information is present in the mongod log file. However, Ops Manager should generate a separate alert for this unique and important state. Without this alert, it is not immediately clear when a user needs take action to bring a replica set back to a healthy state.

What is the problem that needs to be solved? Monitoring is down and Backup is down alerts needs to be sent for each individual MongoDB Agent (Monitoring/Backup Module) which become down, Monitoring is down and Backup is down alerts should include hostname information in them.

Why is it a problem? (the pain) Customer can't easily identify which MongoDB Agent (Monitoring/Backup Module) become down without hostname information available in the alert (in multi-project environment this becomes operational pain for the customer).

In previous versions of Ops Manager, you could filter the process list page. It would be nice to bring that back, so we could quickly identify processes which do not have recent pings, etc.

It would be good to replicate the integration with Slack but using Mattermost. Mattermost is open source and allows the companies to deploy their own messaging server. Some organizations are moving from Slack to Mattermost for this reason.

Add an alert type for outdated agents

Accessing the Real Time metrics tab requires at least the Project Monitoring Admin role but this role has other privileges to administer alerts and manage hosts as well.

It is more appropriate to enable the read-only access user (Project Read Only role) to access the Real Time metrics tab.

Add to The Get Host, Process, System Measurements Ops Manager API the hostInfo.system.memLimitMB.
Or add an API to get the pings value.

We should have a button to sort by shard_name on top

List shards in Deployment > Metrics' shard list in alphabetical order in Cloud Manager UI.