Cloud Management Portal
The Cloud Management Portal (CMP) is the self-service console for Epicor Cloud (SaaS) tenants. Jobs that used to need a support case, such as restarting the application server, copying Live into Pilot or regenerating the data model after adding a UD field, you can do yourself in a few minutes.
Get access
Section titled “Get access”CMP sign-in goes through Epicor Identity Provider (IdP), and your Epicor user has to be linked to it.
- Ask Epicor’s cloud operations team for an IdP account, giving the email address you want to use.
- Log in to your Live environment. Access is set up from Live even if you mainly want to manage Pilot.
- In User Account Security Maintenance, open your user and:
- set Email to the same address as the IdP account,
- set External Identity to that same address,
- tick Security Manager, and
- add the
ECMP_SelfSignUpsecurity group.
- Save, then open System Management > Cloud Management Portal (or search the menu for “cloud m”). Sign in to IdP if prompted and accept the terms the first time.
The first users become tenant administrators, who can add other portal users and assign them roles.
Tenant dashboard and instances
Section titled “Tenant dashboard and instances”The Tenant Dashboard lists every environment you have: Live (production), Pilot and any additional ones such as Third or Education. Epicor calls each one a tenant instance: an application server, its database and the services attached to it, such as Enterprise Search and Data Discovery.
Choose More Details on an instance (or pick it from Tenant Instance) to manage it. Administrators can give instances friendly names with Change Instance Name, which helps when several people manage several environments.
Summary tab: links and site actions
Section titled “Summary tab: links and site actions”The Summary tab gathers the details you otherwise hunt for:
- the web client, server, web MES, REST API and System Agent links for the environment, and a desktop client download link,
- the Kinetic version and build, and
- the Host OS, which tells you whether the environment runs on Windows or Linux (see Linux containers in Epicor Cloud).
It also has the actions you’ll use most:
| Action | Use it when |
|---|---|
| Start / Stop / Restart Site | The environment is unresponsive, BPM or function editors hang, or Support asks for a restart. Restarting disconnects every user of that instance. |
| Start / Stop Task Agent | Scheduled tasks aren’t starting or are stuck. See Scheduled tasks. |
Database tab
Section titled “Database tab”Refresh DB copies a database backup from one instance into another, most often Live into Pilot so
you can test against current data. Pick the source instance and backup date, type confirm and run it.
- You can’t refresh into Live.
- Source and target must be on the same Kinetic version, so refresh before one of them is upgraded, or after both are.
- Everything in the target is overwritten, including customizations, BPMs and layers built there that aren’t in Live yet. Export them with Solution Workbench first.
Data Model Regen regenerates the data model after you add UD columns or tables. You don’t need to stop anything first: the pipeline stops the instance, regenerates and restarts it. Users are disconnected, so run it out of hours. See UD fields.
Upgrade tab
Section titled “Upgrade tab”The Update Policy is either Epicor Managed (Epicor applies updates on its schedule) or Self Managed (you choose when). Changing the policy on Live changes it for every instance in the tenant.
With self-managed updates, choose Update, pick the target version, type confirm and run it. If you
don’t update before Epicor’s deadline, Epicor updates the instance for you.
Check what happened
Section titled “Check what happened”Deployment Status Detail shows every action run against an instance: who ran it, when, its status (pending, complete or cancelled), each step, and a log for steps that have one. Check it after any refresh, regeneration or update, and before telling users the environment is ready.
Good habits
Section titled “Good habits”- Tell users before restarting a site or regenerating the data model; both drop active sessions.
- After refreshing Pilot from Live, change Pilot’s company color back, and disable anything in Pilot that sends email or talks to live external systems.
- Keep at least two tenant administrators so you’re never locked out.