Run a recovery drill before you need a recovery
Every community owner believes their backups work. The ones who have actually restored one into a test server know it. A forty-minute rehearsal turns a guess into a fact you can plan around.
Short answer
Choose the snapshot you'd trust in a real incident, restore it with RestoreCord into a private spare server, compare roles, overwrites and history against the live server, then log the gaps, pin the snapshot and book the next drill.
Why a schedule isn't a plan
On Business and Enterprise, RestoreCord can take backups daily, weekly or monthly. That's a big step up from remembering to click a button, but a schedule only proves that snapshots exist. It never proves they bring back the server you think they do, and RestoreCord's own documentation is clear that restore testing is a separate job.
A drill answers the questions an owner actually gets asked during an incident: how long until we're back, what will be missing, and who does what. You want those answers on paper before the night they matter.
Prepare the drill
- Create a private Discord server you own and name it clearly, for example nightshift-drill.
- Invite your RestoreCord bot to it and drag the bot's role above everything it has to create.
- Open the Backup Previewer on your candidate snapshot. It's on every plan and shows channels and messages before you commit.
- Tell your staff a drill is running so nobody panics at the test server appearing in their list.
Snapshot captured before the permission redesign.
Roles and overwrites look right in the preview.
Backup completed — 42 channels, 18 roles, 6,300 messages.
Rehearse with the snapshot you'd really use, not the newest one by default. In a real attack the right recovery point is the last one created before the first malicious change.
Restore and inspect
Restore the snapshot into the spare server, choosing the message count you'd pick in an emergency. Structure comes back first, then history. While it runs, don't launch a second restore; note any permission failures or skipped objects instead.
When it finishes, walk the result against the live server:
| Check | What good looks like | Common gap |
|---|---|---|
| Roles | Same names, order and colours | Bot role sat too low to create some of them |
| Permission overwrites | Private channels are still private | A category synced permissions differently |
| Message history | Recent messages present up to your plan's per-channel cap | Excluded or unreadable channels are empty |
| Attachments | Images present on Business, every file type on Enterprise | Files over 8 MB are skipped |
| Emojis | Custom emojis restored (Enterprise) | Emoji capture was off when the snapshot was taken |
The parts a snapshot can't rehearse
Members are not stored inside a snapshot. In a real recovery they come back through a separate pull, and only people who authorized through RestoreCord before the incident are eligible. Their roles return as a third step: after a restore, RestoreCord keeps each member's role mapping for 7 days and offers Restore Roles from a dashboard banner once they're back in the server.
- +Categories and channels
- +Roles and permission overwrites
- +Server settings
- +Messages (plan-limited per channel)
- +Attachments (Business and Enterprise)
- +Emojis (optional toggle)
- −Members — pulled separately
- −Role assignments — restored separately
- −Voice/message activity and presence
- −Files over 8 MB
- −Channels the bot cannot read
- −Excluded channels you deselected
You don't need to pull your live community into a test server to prove this works. What you should check is the input: is verification switched on, and is the share of members who have verified growing? That number is your ceiling on recovery.
Write the drill down
Keep a short log in your staff channel or wiki. The format matters less than doing it every time:
Recovery drill: Night Shift, Q4
Snapshot used ........ Pre-season (pinned)
Restored into ........ nightshift-drill (private)
Started / finished ... 14:02 / 14:31 UTC
Roles and channels ... match, 1 overwrite fixed by hand
Private channels ..... still private
Messages ............. history present up to the plan cap
Gaps found ........... bot role was below "Moderator"
Owner sign-off ....... yes, next drill booked for Q1If the snapshot passed, pin it. At the plan limit, a scheduled backup rotates out the oldest unpinned snapshot for that server, and pinned ones are never removed, so a pinned, tested recovery point always survives. Leave unpinned room for the schedule to keep working.
Delete the drill server afterwards, or keep it locked and empty for the next rehearsal. Don't let it turn into a second community nobody moderates.
A drill is also a good moment to check who on your staff can restore. That's the Manage Backups permission; see which permissions each role should hold. More on running the community like a business on the RestoreCord for Owners home page.
The drill in four moves
Pick the recovery point
Open the snapshot you would trust in a real incident and check its channels in the Backup Previewer.
Restore into a spare server
Invite your bot to a private test server, place its role at the top and restore the snapshot there.
Inspect what came back
Compare roles, permission overwrites, private channels and message history against the live server.
Log it and book the next one
Write down timings and gaps, pin the snapshot if it passed, and put the next drill in the calendar.
Before you sign the drill off
Tick each one in your log. Anything left open becomes a task for the next drill.
Owner accounts use 2FA
Your RestoreCord login and the bot owner's Discord account both have two-factor on.
The bot role sits on top
It's above every role the restore had to create in the test server.
A tested snapshot is pinned
Rotation can't remove the recovery point you just proved.
Restore rights are narrow
Only the owner or one deputy holds Manage Backups.
Verification is on
New members authorize as they join, so the pullable list keeps growing.
The next drill is booked
A date in the calendar, with a named person running it.
Recovery drill questions
Know your server comes back
Snapshots on every plan, schedules on Business and Enterprise: rehearse once with RestoreCord and stop guessing.