Slate BETA 2.3.0 · for IT

Slate for IT

Everything IT does with Slate, from the first install to the day you move the server. Each step is one thing to do. The words in bold are the words on the screen.

What the studio is made of

Slate has three parts. One server PC runs Slate Server and keeps the studio database on its own disk. One shared folder on the network, called Slate_Central, holds files every PC needs. Every other PC is a workstation with Slate Studio, Slate Operations, or both.

Workstations save and read through the server. Files go to the shared folder. The server PC also needs the shared folder: it publishes updates there.

Two folders called Slate_CentralThe shared Slate_Central is on the network, for example Z:\Slate_Central or \\server\share\Slate_Central. The server PC also has a local folder with the same name, %LOCALAPPDATA%\Slate_Central. That local one holds the database and the Recovery Key's fingerprint. The server's backups go to the shared one, in Slate_Central\Slate_Server\Backups. Never mix them up, and never delete the local one.

Three installers: who gets which

Installer	Start menu	Holds
`setup_Slate Server_vBETA 2.3.0.exe`	Slate Server, Recover Slate	Slate Server, the database (PostgreSQL 17), the connection pool (PgBouncer), Recover Slate
`setup_Slate_Studio_vBETA 2.3.0.exe`	Slate	Slate Studio (window title Slate VFX): the PRODUCTION screens, ffmpeg, OpenRV
`setup_Slate_Ops_vBETA 2.3.0.exe`	Slate Operations	Slate Operations: Attendance, Leave, IT Support and the office screens, ffmpeg

Slate Studio shows only Home, the PRODUCTION screens and Settings. Everything under PEOPLE (Attendance, Leave, Joining & Leaving), IT & INFRA (Hardware, Licences, IT Support, Deployment) and ADMINISTRATION (Users & Roles, Admin Panel) is only in Slate Operations. Punching in, asking for leave and raising an IT ticket all happen in Slate Operations. So:

Kind of PC	Install	Why
Artist, lead, coordinator	Studio + Operations	Studio for the shot work; Operations to punch in, ask for leave and raise tickets
Supervisor (approves leave)	Studio + Operations	Leave approval, team attendance and the Admin Panel are only in Operations
HR, IT, production office	Operations	Everything they need is in Operations. Exception: anyone in the office who works on the VFX Dashboard, Scheduling or Bidding also needs Studio, because those screens are only there.
The server PC	Server only	Nobody works on it. Keep it for the database.

Both on one PCThe two workstation programs live side by side and share the same settings on that PC (%LOCALAPPDATA%\Slate). Install both, answer the shared-folder question the same way, and the second one already knows the server.

All three installers install for the signed-in Windows user, with no administrator rights, into %LOCALAPPDATA%\Programs\. Each one starts its program when Windows starts. Nothing needs the internet.

Setting up a brand-new studio

Do these in this order. The order matters: the server must exist before any workstation can sign in, and the first Slate that opens builds the studio's tables and roles.

Green steps happen on the server PC. Step 6 is on a workstation. Step 8 is in Slate Operations on any PC.

1. Install Slate Server

Pick the server PCAlways on, wired network, a fixed address (a DHCP reservation). The database lives on its own disk.
Run the server installersetup_Slate Server_vBETA 2.3.0.exe. Sign in to Windows as the account that will run the server every day.
Select Slate_Central Shared FolderType the shared folder in Server Root Path, e.g. Z:\Slate_Central. It cannot be left blank. Use the same path on every PC.
FinishThe Start menu now has Slate Server and Recover Slate.

"This folder does not exist on this machine"The installer asks Use it anyway? when it cannot see the shared folder. Answer Yes only if the drive is simply not mapped yet. If it is a typo, answer No and fix it. A wrong path means Slate works on that PC only and shares nothing.

2. First start: where the database lives

Open Slate Server. It opens on the Dashboard. If you want the database somewhere other than the default %LOCALAPPDATA%\Slate_Central\LocalDatabase (for example a second internal drive), go to Settings first, type it in Database folder, and press Save Configuration. Keep it on this PC's own disk: Recover Slate refuses a database on a network drive.

Brand-new studio: Create a new empty database here. A studio that already has data: Use an existing database folder... and pick the folder that holds the file PG_VERSION. A folder without it is refused with "Not a database".

Database Server PowerTurn the switch on. The badge says Starting..., then Server Online.
Read SYSTEM LOGSA new studio shows: "This studio's own database password was made and is kept on this PC only."
Tables in Database: 0Normal now. The warning "exists but has no tables" goes away when the first workstation opens Slate.

3. The Recovery Key

Right after the database starts for the first time, Slate Server shows Your Slate Recovery Key. It looks like XXXXX-XXXXX-XXXXX-XXXXX-XXXXX. It is shown once. Slate keeps only a one-way fingerprint of it, so nobody can read it back later.

The key above is an example. Close stays grey until you tick the box. Esc and the window's cross do nothing until then.

Never lose the Recovery KeyIt is the only way back in when nobody can sign in. If it is lost, a Windows administrator on the server PC can make a new one: right-click Recover Slate, Run as administrator, leave the key box empty and press Make a new Recovery Key. The old key stops working at once. Without the key and without Windows administrator rights on the server PC, there is no way back.

4. Open the firewall

Settings → Allow FirewallWindows asks for administrator rights. Say yes.
Check the log"Firewall rules are in place; workstations can connect." Beside the buttons: Firewall allowed.
Firewall not changedThe request was refused. Press it again, or add the three rules by hand.

Windows firewall rule	Port	Used for
Slate Central Server (Database)	TCP 5440	The database (the PostgreSQL port in Settings)
Slate Central Server (Pool)	TCP 6432	The connection pool (the Connection pool port). Without it every workstation waits before it falls back to 5440.
Slate Central Server (Discovery)	UDP 54320	Find server on the workstations

No installer opens the firewall. Only this button does. If you change a port in Settings, press Allow Firewall again.

Steps 5 to 8 have their own pages: The studio's database password and The first administrator, people and roles.

The studio's database password

Every workstation signs in to the database with one shared account, ut_vfx_app, and one password. People's own Slate passwords are separate: this one is only between the PCs and the server.

Each studio has its own. Slate Server makes a strong random one when it creates a new database. No password ships with Slate any more.
On the server PC it is kept encrypted by Windows, in slate_recovery\db_secrets.dat beside the database. Only this PC can read that file.
On a workstation it is kept in Windows Credential Manager (a generic credential named Slate, user db_password), not in a settings file.
You read it in Recover Slate: Database passwords → Show app password. Write it down and keep it with the Recovery Key. You need it again if you ever move the server.

Leave Port, Database name and Database user as they are unless you changed them on the server. Find server fills in the address when the discovery port is open. Later changes use Reconfigure server / database on the sign-in screen, which asks the same things.

A database that still uses the old shared password

Every Slate before 2.2.0 shipped with the same database password, and it is public. When Slate Server 2.2.0 starts on such a database, it keeps that old password working for now and publishes a new one of the studio's own. The log says "This studio still uses the database password older Slate versions shipped with." The health check in Recover Slate warns Studio password until you switch.

Recommended first step when no PC is running Slate yetIf your server already has its database from an older Slate, and no workstation runs Slate yet, switch to the studio's own password now, before you set up any workstation. Nobody is using the old one, so nobody can be cut off, and every workstation you set up afterwards gets the new password from the start.

All five steps are on the server PC, in Slate Server's Recovery page or the Recover Slate Start-menu item. Both are the same page.

Then set up the workstations with the password Show app password now gives. If you ever need to go back, Show app password also prints "Before the last change it was: ...". Type that one in the two password boxes and press Set app password (workstations).

The same switch on a studio where workstations are already running

Do not switch straight away. First get every workstation on 2.2.0 and open Slate on each once: each one learns the published password quietly. Show app password lists the ones that have it, under "Workstations that have it". When every PC is listed, press Switch to the published password at a quiet moment. Anyone who had Slate open may need to restart it. A PC that missed it says it has no password: use Reconfigure server / database on its sign-in screen.

Do not use Settings to change the passwordTyping a new password into Slate Server's Settings → Database password and saving changes it for the server at once, and cuts off every workstation that does not have it. The gentle way is always Recover Slate: Publish, then Switch.

The first administrator, people and roles

The first workstation

Install on your own PCSlate Operations (and Slate Studio if you want it). The installer asks for the same Server Root Path as the server.
Set up Slate on this computerOpens at the first start. Find server, type the database password, Test connection, Save.
The sign-in window appearsSlate has now built the studio's tables and its starting roles. On the server, Tables in Database is above 0.

Make your administrator

A new database starts with one account, admin, with the Developer role (full access), while the switch no_default_accounts is off (it starts off). Its starter password is the well-known admin123, so Slate marks it: the first sign-in must choose a new password before anything opens. The same happens if Slate ever brings admin back after it was removed. Sign in as admin yourself, straight away, before you set up any other PC:

User name adminOn your workstation's sign-in window. Password admin123. Press Sign in.
Choose your passwordOpens by itself. Your own password twice, 8 or more characters. Save password.
Keep it with the keyWrite the admin password down with the Recovery Key. Cancel on that window only goes back to sign-in.
Create peopleIn Users & Roles, below. Give yourself a personal account too, and keep admin for emergencies.

The other way in: Recover SlateOn the server PC: Recovery, type the Recovery Key, Unlock. In Accounts type a username (blank means admin) and a new password twice, then press Create or restore administrator. The output says "...is an active administrator again, with the new password." or "Created the administrator ... with the new password." That person chooses their own password at the next sign-in. This works at any time, even before any workstation has opened Slate: a new studio always gets its starting roles (Supervisor, Lead, Coordinator, Artist and the rest), whoever made the first account.

With no_default_accounts onNo starter account is ever made or brought back. The first-run notice then says "To make the first administrator, open Recover Slate on the server PC and use Accounts > Create or restore administrator." Type any username there; Slate creates it with the Developer role.

Users & Roles

In Slate Operations, under ADMINISTRATION. It has two tabs. Users needs the ability Manage users (Admin, IT, HR, Developer). Roles & Permissions needs Edit roles and permissions (Admin, IT, HR, Developer). The standard IT role has both, so IT creates accounts, resets passwords and deactivates. A studio upgraded from an older Slate gets this once, on the first start.

The Users tab. The Location column sits between Reports to and Status; it is left out here for room.

To do this	Press	What to know
Add one person	Add user	Username (2 to 40 letters, numbers, dot, dash or underscore, e.g. `priya.sharma`), Display name, Department, Roles (at least one), a first Password, Reports to. They choose their own password at the first sign-in.
Add many people	Import...	Excel or CSV. Download template first. Pick a Role for everyone and a First password (8 or more characters). Admin, Developer and roles with sensitive abilities are never offered here.
Someone forgot their password	Reset password	8 characters or more. They must choose their own at the next sign-in. Nobody is ever locked out for wrong tries: Slate only waits one second.
Someone leaves	Deactivate	History is kept. Refused while they still have leave waiting or machines issued: handle those first. A manager with a team gets a New manager picker.
Someone comes back	Reactivate	Offers to clear a last working day that has passed.
Remove a mistake	Delete...	Only for an account with no history (no attendance, leave, machines, checklists or tickets). Everyone else: Deactivate.

The last administrator is protectedSlate refuses any change that would leave no active administrator: "This would leave Slate with no active administrator, and then nobody could sign in to put it right." It also refuses to leave nobody who can manage users. admin is a system account: it cannot be deleted or deactivated.

Roles: what each one opens

The table shows the roles 2.2.0 starts with. A studio upgraded from an older Slate gets these additions once, on the first start: IT gets Users & Roles (Manage users), the Admin Panel and Message and restart workstations; Production Coordinator and Producer get Bidding and Edit bids. Only roles with exactly those names are changed, nothing is taken away, and a role the studio changes afterwards keeps its choice. A role decides which screens appear in the sidebar and what the person may do there. A screen the role cannot open is simply not in the list. Home, Attendance, Leave and IT Support are always open to everyone (in Slate Operations).

Role	Opens	Notable abilities
Developer	Everything, including Tester Panel	Full access. Cannot be changed or deleted. Recover Slate makes administrators with this role.
Admin	Every screen except Tester Panel	Every ability except the own-department limit, being given shots and destructive test tools.
IT	IT & Infra (Hardware, Licences, IT queue, Deployment, IT side of Joining & Leaving), Users & Roles, Admin Panel (Live Ops), Settings	Work the IT desk, Manage users, Edit roles and permissions, Message and restart workstations, Studio settings, See licences
HR	HR (people, leave, joining), Settings	HR leave stage, Manage users, See team attendance, Edit roles and permissions
Production Head	VFX Dashboard, Timeline Viewer, Scheduling, Bidding, Stock Viewer, Build & Ingest, CAP Rename, Settings	Edit the dashboard and schedule, Edit and Approve bids, Approve leave, See team attendance, See licences
Production Coordinator, Producer	As Production Head	Edit the dashboard and schedule, Edit bids, See every shot
Supervisor	Build & Ingest, CAP Rename, Stock Viewer, Timeline Viewer, VFX Dashboard, Settings, Admin Panel	Approve leave (first stage), Force-save over others. Admin Panel is Live Ops, read-only.
Comp Supervisor, Roto Prep Supervisor	VFX Dashboard, Timeline Viewer, Stock Viewer, CAP Rename, Settings	Supervisor abilities, limited to their own department
Coordinator	Build & Ingest, VFX Dashboard, Stock Viewer, CAP Rename, Timeline Viewer, Settings	–
Lead, Team Lead	VFX Dashboard, Timeline Viewer, Stock Viewer, CAP Rename, Settings	Approve leave (first stage), own department; can be given shots
Editor	Timeline Viewer, Stock Viewer, VFX Dashboard, CAP Rename, Settings	See every shot
Artist, Compositor, Roto Artist, Paint Artist, Deage Artist, AI Artist, DMP, CG	VFX Dashboard, Stock Viewer, CAP Rename, Settings	Set own shot status, Can be given shots
Tester	Build & Ingest, CAP Rename, Stock Viewer, Tester Panel, Settings	For testing Slate. Give it to nobody else.

Abilities: WHAT THIS ROLE CAN DO

Open Roles & Permissions, pick a role on the left, then tick under TABS THIS ROLE CAN OPEN and WHAT THIS ROLE CAN DO, and press Save changes. A change shows the next time that person signs in. Some abilities come with a role's name: they show ticked and greyed out.

Ability	Plain meaning	Ability	Plain meaning
Edit the dashboard	Change shots on the VFX Dashboard	Can be given shots	Appears in the Artist pickers
…only their own department	Limits the one before to their department	See every shot	All shots of a project, not only their own
Set own shot status	Move their own shots along	Edit the schedule	Change Scheduling
Approve leave (first stage)	The supervisor's yes or no	Edit bids / Approve bids	Make bids / decide them. Approving bids also allows Bidding settings: Days per shot, margins, day rates, Tax on rupee bids (GST) and the Studio name
HR leave stage	The final, HR yes or no	Archive or delete projects	Archive, and Delete permanently
Work the IT desk	The IT queue, and changes in Hardware, Licences, Deployment and the IT checklist	See licences	Read Licences, get renewal reminders
Manage users	The Users tab	System administration	Everything in the Admin Panel, including Audit Logs and Data Center
Edit roles and permissions	The Roles & Permissions tab	Studio settings	Studio-wide cards in Settings, such as working hours and the licence renewal warning
See team attendance	Everyone's attendance	Destructive test tools	Large Tester Panel runs
Ingest stock	Add to and delete from the stock library	Force-save over others	Keep their version when two people save one shot
Excel export	Export to the project's Excel backup	Message and restart workstations	Live Ops broadcast, Restart, Shut down, Remove from Live Ops. Not Audit Logs or Data Center
		Full access (tick box)	Every tab and ability, including future ones

Only Admin and Developer hand out the big onesFull access, Manage users, Edit roles and permissions, System administration, Message and restart workstations, Studio settings, Archive or delete projects and Destructive test tools can be given only by Admin or Developer. Everyone else can hand on only what they have themselves.

Looking after the server

Slate Server has five pages in its left-hand list: Dashboard, Analytics, Operations, Settings and Recovery. Keep the window open, or in the tray. Closing it asks Stop the studio database?: Keep running in the tray closes only the window; Stop server takes every workstation offline.

A healthy server: Server Online, tables above 0, the pool on port 6432 and no warning lines under the cards. The numbers here are examples.

Every morning: two minutes

Look at	Good	If not
Badge and power switch	Server Online, switch on	Turn it on. If it fails, read the box "The database did not start".
Tables in Database	Above 0	0 on an old studio means the wrong folder. See When something goes wrong.
Connection Pool	`port 6432`	stopped or wrong database: press Restart pool.
Warning lines under the cards	None	Read them. Each says what to do.
Operations: last backup	"Last backup 20 min ago - kept in …\Slate_Server\Backups"	Red text: press Back up now and find out why the automatic one failed (SYSTEM LOGS). If it says the studio folder cannot be reached, the backups are on the server PC only until the share is back.

Analytics

Database Analytics shows Active PCs, Total Projects, Stock Assets, DB Load, Database Size, CPU Usage and RAM Usage, then Live Connected Clients. When one PC holds the database up and everyone is slow, select its session and press Disconnect selected. Slate asks the session to stop first, and cuts it off only if it ignores that. Anything that PC was part way through saving is lost.

Operations: backups

Slate Server backs the database up every hour by itself, while it runs (window open or in the tray) and the database is on. It looks every 10 minutes and takes a backup once the newest is an hour old. The backups go to the shared folder, in Slate_Central\Slate_Server\Backups, so they survive the server PC. Kept: the newest 24, plus the newest of each day for 14 days; older ones are deleted by themselves.

If the shared folder cannot be reached, that hour's backup goes to the Backups folder beside the database on the server PC, the status bar and the log say so in red, and the next one tries the shared folder again. A server that keeps a backup_dir in slate_server_config.json uses that folder instead.

Make Slate_Server admins-onlyThe backups hold everything in the database, including password hashes and HR records. On the file server, give Slate_Central\Slate_Server to IT and the server PC's account only, not to everyone who can open Slate_Central.

Button	What it does
Back up now	A full copy of the database, now. Do it before any upgrade, restore or password change.
Keep 30 days, always keep 7	A tidy-up rule for Delete old backups only. The hourly backups keep to their own rule (above).
Delete old backups	Applies that rule now. It lists what it will delete and asks first. It cannot be undone.
Restore from file…	Replaces everything in the live database with a `.dump` file. Close every workstation first. You must type `RESTORE`.

Backups are files named like slate_ut_vfx_20261006_141800.dump. Make sure the file server itself is backed up: it now holds the studio's only off-PC copy.

Operations: maintenance, diagnostics, the log

Job (Every)	Runs	Button
Vacuum and analyze (every 7 days)	By hand	Clean up the database (vacuum). Queries get slower without it.
Rebuild indexes (every 30 days)	By hand	Rebuild indexes. Also after a big ingest or a restore.
Backup (every hour)	By itself, while Slate Server runs	Back up now
Credit comp off (every day)	By itself: about 5 minutes after Slate Server opens, then daily	Credit comp off. A second run the same day does nothing more. Runs only while Slate Server is open (window or tray).

The Last run column says "1 day late", "5 days late" or "Never run". Run diagnostics checks the configuration, the database, the server, the pool and the update folder, and lists what is wrong in plain words. Run it first on any support call. Server log: Open log opens pg_server.log; Trim log keeps the last 2 MB. Danger zone: Force kill the database… only when the power switch does not respond. It stops every PostgreSQL on the PC at once.

Settings

Server Configuration: Database folder, PostgreSQL port (5440), Connection pool port (6432), Database name (ut_vfx), Database password, Most connections (100). Then Allow Firewall and Save Configuration. While the server runs, new folders and ports apply only after you turn the power off and on. If the note says NOT saved, nothing changed: read why.

Do not change the Database nameWorkstations ask for ut_vfx. The database answers an unknown name by making a new empty one, and the studio's work then seems to vanish.

Publishing updates

From 2.2.0, the update is the installer itself. You install Slate by hand once. After that, you publish the new installers on Slate Server and every workstation updates itself from the shared folder.

Nothing goes to the internet except your own download in step 1. The installer runs silently on each PC and keeps that PC's settings.

Publish an update, step by step

DownloadThe new setup_Slate_Studio_v….exe and setup_Slate_Ops_v….exe from the owner's GitHub release, onto the server PC.
Back up nowOperations. A new version may change the database the first time a workstation opens it.
Required?Tick Required (installs even if the artist chose later) before publishing, if nobody may put it off.
Publish update…Settings. Choose one or both installers. Any other file name is refused.
Read the result"Slate Studio BETA … is now offered to every Slate Studio workstation." and the folder.

Slate Server copies each installer into Slate_Central\Updates\releases, checks the copy, then writes manifest_studio.json or manifest_ops.json beside it, then deletes that program's older installers. Installers are hundreds of MB, so this takes a while; the line above the buttons shows each step.

What the artist sees

Within 5 minutes Slate finds the update (10 seconds after start, then every 5 minutes). To spare the network, each PC first waits a random time of up to 10 minutes, then copies the installer in the background.
A new version of Slate is ready. Update now asks about unsaved work, closes Slate, installs and opens Slate again. When I close Slate installs it at the next real close (not a sign-out); Slate then does not reopen by itself.
If an update is made required while the ordinary offer is still open on a PC, the countdown replaces that offer. The silent installer starts only once Slate has fully closed.
Settings → Check for updates looks at once, without the wait, and always says what it found.
A copy whose checksum does not match is deleted and never installed: "...is not the file that was published (its checksum does not match)...".

Required updates, rollback, the server itself

You want	Do this
Everybody on it now	Press Require the current update now, answer Yes. PCs that chose "later" get Slate must update with a 5-minute countdown at their next check (within 5 minutes). Then Slate closes, updates and opens again.
Go back to the previous version	Publish the older installer the same way. Any version different from the running one is offered, so an older one is a rollback. Keep the last good installers on the server PC for this.
Update Slate Server	Update this server…, choose `setup_Slate Server_v….exe`, answer Yes. It says how many workstations are connected. The installer stops the database while it works; workstations reconnect by themselves. Database, backups and settings are kept. Do it at a quiet time.
Slate will not start on one PC	On that PC, run the installer from `Slate_Central\Updates\releases` by hand. It keeps the PC's settings.

Updating to 2.2.0 itselfPCs on 2.0.32 still have the old update system. Install 2.2.0 by hand on every PC. Self-updating starts with the next version you publish. Try the first published update on one spare PC before relying on it.

Recover Slate and the security switches

Recover Slate is for when nobody can sign in, or a security change went wrong. It works only on the server PC, and only with the Recovery Key. It never needs a Slate password or a database password.

Open it from the Start menu (Recover Slate), or click Recovery in Slate Server's left-hand list. Both show the same page. The Start-menu one works even when the Slate Server window will not start. It refuses to run on any other PC, or against a database on a network drive.

The health check runs without the key and changes nothing. Everything below Unlock needs the key. Unlocking lasts ten minutes. Wrong keys make you wait (5 seconds, growing to 5 minutes at most), never a permanent lock.

Every action, and when to use it

Panel	Button	Use it when
Health check	Check now	Always first, and again after every fix. Lines such as This PC, Server login, Workstation login, Administrator, Studio password, Clock, Recovery Key.
	Stop leftover pool	"A connection pool is still running from an earlier start." Or just start Slate Server.
	Close an interrupted repair	"An interrupted recovery left the access rules open." Starting Slate Server also closes it.
Accounts	Reset password and switch the account back on	Someone forgot their password, or was switched off or given a past last day by mistake. Also clears that last day. They choose their own password at the next sign-in.
	Create or restore administrator	No administrator is left, or a new studio needs its first. Makes or mends the account with the Developer role and gives Developer full access again.
Database passwords	Show app password	To set up a workstation, or to see who has learned a published password. Also shows the one before the last change.
	Publish a new app password, then Switch to the published password	To change the workstations' password without cutting anyone off. See the password page.
	Set app password (workstations)	Changes it at once. Use it to put back the password the workstations already have, for example after moving the server. Then press Restart pool on the Dashboard.
	Set superuser password (this server)	The server says it cannot log in to its own database. Workstations are not affected. Keep this password with the Recovery Key.
Switches and snapshots	Turn off / Log only / Turn on	See the switches below.
	Restore the last snapshot	A settings file or the access rules got damaged. Tick Also put the users and roles back only if accounts were damaged too. The current files are kept as a snapshot first.
	Make a new Recovery Key	The key may have been seen by the wrong person, or is lost (then run as a Windows administrator, key box empty).

The never-lock-out rules Every switch starts Off. A snapshot is taken before every change and every repair, so each can be undone. Turn on first checks that people can still get in, refuses if not, and undoes itself if the check afterwards fails. Turn off works even when the database is down. Slate never removes, demotes or switches off the last administrator. A wrong Slate password only waits one second; nobody is ever locked out for wrong tries. Every action is written to slate_recovery\recovery.log, never with a password or the key.

The security switches

Choose one switch in the list, read the line under it (what it does and its state now), then press a button. Switches are turned on one at a time. Every security switch allows Turn off only.

Order	Switch	When on, in plain words	Before you turn it on
1	`split_superuser_password`	The database's master account gets its own password, kept on the server PC only. Workstations keep theirs.	Nothing. If Recover Slate says the pool did not pick it up, press Restart pool.
2	`no_default_accounts`	The starter `admin` account is never created or brought back. The first-run notice points to Recover Slate instead.	You have your own administrator, and the Recovery Key is safe.
3	`no_plaintext_passwords`	Passwords stored in old, unsafe formats no longer open an account.	Let everybody sign in once: each sign-in re-stores the password safely. It is refused while an active account still has one.
4	`refuse_inactive_signin`	Deactivated people, and people past their last working day, cannot sign in. `admin` and `developer` are always let in.	Check Users & Roles for anyone wrongly marked as left.
5	`strict_pg_hba`	The master account connects only from the server PC. Workstations reach only the studio database.	Refused while a workstation is connected as the master account. Try Log only first.
6	`pgbouncer_hba`	The pool's admin console works from the server PC only.	Refused while someone uses that console from another PC.
7	`signed_fleet_commands`	Messages, restarts and shut-downs carry a signature; workstations ignore any without one.	Everyone who sends messages or restarts (admins, and IT with Message and restart workstations) on 2.2.0. Log only for a day, then look for "would be refused" in the workstations' logs.
8	`hide_password_hashes`	Password fingerprints move where workstations cannot read them; the database checks passwords itself.	Every workstation on 2.2.0 first. A workstation cut off from the server cannot sign anybody in from its local copy.
9	`no_sqlite_fallback`	When the server is down, workstations do not open their local copy (which has its own starter admin account).	With it on, nobody can work while the server is down. Last, and only if that is acceptable.

The order above starts with the switches that cannot affect anyone's work and ends with the ones that can. Turn on one, work a normal day, check, then the next.

Your screens in Slate Operations

Under IT & INFRA: IT Support, Hardware, Licences and Deployment. Under PEOPLE: your half of Joining & Leaving. Changing anything here needs the ability Work the IT desk. Without it, Hardware, Licences and Deployment are read-only.

IT Support: the queue

With Work the IT desk you get two pages: Queue (everyone's tickets) and My tickets (your own). The queue is sorted by what is closest to breaching, not by what is newest. Five figures across the top work as filters: Unresolved, Breached, At risk, Unassigned, Mine. The sidebar badge counts the unassigned ones.

Nobody picks P1 by hand: the person raising the ticket answers Who is affected and Can you carry on, and the grid gives the priority. New P1 and P2 tickets notify IT.

Button	What it does
Assign to me	Takes the ticket. It moves to In Progress.
Set status…	Waiting, Resolved and Closed all need a note, and the person who raised it reads that note. Waiting needs "What do you need from them?".
More ▾	Assign to…, Unassign, Change priority… (needs a reason; raising it restarts the clock from now), Responded by phone (stops the reply clock when you answered off-screen).
New ticket	Log a problem for somebody else: a phone call or a walk-up.
SLA report, Export…	How many promises were kept, per priority, for the last 12 months or one month.

Working time is the Working hours in Settings → Studio currency, rates and hours (10:00 to 19:00 unless changed), on the studio's working days from Studio policy, skipping holidays. A day means one standard working day (9 hours unless changed). So a ticket raised on Saturday evening is not breached on Monday morning. At risk means the last quarter of the reply time, or the last fifth of the fix time. Inside a ticket, tick Internal note (IT only) for words the requester must not see.

Hardware

Machines the studio owns, who has them, and their state. Figures: Machines, Available, In repair, Out on loan. Buttons: Add machine, Edit, Issue to…, Collect, History, Delete, Sync from Live Ops, Export….

Status	Means
Available	Free to issue.
Active	Issued to someone. Shown as In service in the Edit window.
Repair	Never offered when issuing. Collecting it back does not clear Repair.
Retired, Lost, Disposed	Hidden unless you tick Show retired. A machine that was ever issued is retired, not deleted: Delete is only for a machine added by mistake.

The Warranty column turns amber within 60 days of the end and red once expired. History lists every issue, return, status change and rename, with who did it.

Licences

What we bought, what we actually use, and what to do before each renewal. The honest measure of a floating licence is its peak use, so record readings when the studio is busy. Buttons: Readings…, Record usage, Import from licence server… (reads a FlexLM lmstat -a or RLM rlmstat -a text report), Add licence, Edit, Remove, Export…. Choose Peak over 30 days, 90 days or a year.

State	Means	Do
Over-subscribed	More in use than we own	First. This is the one that ends in an audit letter.
Expired	Past its renewal date	Renew or remove it. Anyone relying on it is stuck.
Renews soon	Inside the Licence renewal warning (45 days unless changed in Settings)	Decide before purchasing needs its lead time.
Under-used	Peak below 60% of the seats bought	Seats to drop at renewal.
No readings / Not measured lately	Bought but never, or not recently, measured	Record usage.
Healthy	Nothing to do	–

The bell reminds IT and everyone with See licences when a licence enters the warning window, at 14 days and on the day. The What this means column says it in numbers you can take to purchasing.

Deployment

Deployment log: A record of what was installed where. Slate does not push it. Press Record deployment, type the Package, Version and Machines (several, separated by commas, make one record each), then Record. A new record is Pending; later press Mark success or Mark failed (with an optional reason). Figures: Installs recorded, Success rate, Awaiting outcome. Slate updates published from Slate Server are not logged here by themselves: record them if you want the trail.

Joining & Leaving: IT's half

HR start a joining or leaving checklist; IT never do. Your view is Accounts, access and machines - what is still outstanding, with the cards On your list, Joining, Leaving, and Kit not returned (machines issued to people who have left; shown only when there are some).

Joining: IT lines	Leaving: IT lines
Domain account created Email and calendar set up Workstation issued DCC software installed Project shares mounted Render farm access	Render farm access revoked Project shares unmounted Workstation returned Email forwarded and mailbox archived Domain account disabled plus one "Return <machine>" line per machine still issued

Select the personThe filter starts on Everyone outstanding.
Issue machinePick from the free machines. Hardware shows it as theirs, and Workstation issued ticks itself.
Mark doneFor each other line, when it is really done. Reopen undoes it.
Collect machineOn the way out. Workstation returned ticks once nothing is left out.

Deactivating waits for youHR cannot deactivate an account while machines are still issued to that person. Collect the kit first.

The Admin Panel

In Slate Operations, under ADMINISTRATION. It has three pages: Live Ops, Audit Logs and Data Center. It needs the studio server: in LOCAL MODE it says "Admin Panel needs the studio server". The standard IT role has it, with Message and restart workstations: broadcasts, restarts, shut-downs and Remove from Live Ops, but not Audit Logs or Data Center. Upgraded studios get this once, on the first start.

Part	Admin, Developer	IT	Supervisor	What it does
Live Ops: the machine cards, Fleet report, Refresh	Yes	Yes	Yes, read-only	Every PC that runs Slate: Online, Not responding, Offline or Unknown. Problems first.
View system specs	Yes	Yes	Yes	The PC's hardware, from its own report.
Broadcast: Send alert	Yes	Yes	No	A message to every workstation: "Message from <you>" with OK.
Restart… / Shut down…	Yes	Yes	No	One PC at a time, from its card's menu (Actions for this machine or right-click).
Remove from Live Ops…	Yes	Yes	No	For a PC that is gone for good (Offline or Unknown). It comes back if it reports again.
Audit Logs	Yes	No	No	Workstation logs, Change history, Audit trail (every sign-in, success and failure, never a password).
Data Center	Yes	No	No	The database's tables, and a SQL console that is always read-only. Password columns are hidden. People are changed in Users & Roles, never here.

Nobody's PC restarts without them agreeing. Use a broadcast first if many people must save their work.

Live Ops reads a small status file each PC writes to Slate_Central\LiveStatus about every 30 seconds, and judges it by the shared folder's clock, so a PC with a wrong clock does not look online. Messages and restarts travel through Slate_Central\Commands. A PC that cannot reach the shared folder neither shows in Live Ops nor gets commands.

Moving, uninstalling, and where things live

Where things live

What	Where	Safe to delete?
The programs	`%LOCALAPPDATA%\Programs\Slate Server`, `...\Slate Studio`, `...\Slate Operations`	Uninstall instead. A reinstall brings them back.
The database	Server PC: `%LOCALAPPDATA%\Slate_Central\LocalDatabase`, or the Database folder in Settings	No. This is the studio's data.
Backups	Shared folder: `Slate_Central\Slate_Server\Backups` (admins only). When the share was down: `Backups` beside the database folder on the server PC	No. Slate deletes old ones itself.
Where the server is	`Slate_Central\Slate_Server\server.json`: its address, ports and id, rewritten at every start	Yes. The server writes it again.
Recovery folder	Server PC: `slate_recovery`, beside the database folder: the key's fingerprint, the passwords (`db_secrets.dat`), snapshots, `recovery.log`	No. Without it Recover Slate cannot unlock and the server loses its passwords.
Server settings and logs	`%LOCALAPPDATA%\Slate_Central\slate_server_config.json`; `pg_server.log` and `pgbouncer\` beside the database folder	Trim the log from Operations. The pool's files are rewritten at every start; do not edit them.
Workstation settings	`%LOCALAPPDATA%\Slate\config.json`; the database password in Credential Manager	Deleting them makes Slate ask Set up Slate on this computer again.
Workstation logs	`%LOCALAPPDATA%\Slate\Logs` (`latest.log`, `native_crash.log`); a copy in `Slate_Central\Logs\<PC>_<user>.log`	Yes, but send them with any bug report first.
Workstation cache	`%LOCALAPPDATA%\Slate\Cache`	Yes. It is rebuilt.
Downloaded updates	`%TEMP%\SlateUpdate`	Yes. Slate deletes them itself once installed.
Shared files	The shared `Slate_Central`: Updates, Commands, LiveStatus, Logs, Cache, Thumbnails, Config, Attendance, Lineups…	No. Every PC uses it.

Replacing the server PC

Old PC: stop itIf it still works: Back up now, then turn the power off, or take it off the network. Note the app password (Show app password).
New PC: installInstall Slate Server and give it the same shared folder.
RestoreTurn on the power. It finds no database here but finds the backups: press Restore the newest backup from the share (taken … ago).
Give it the passwordPrint the new Recovery Key it shows. Then Recovery, Unlock, type the noted password, Set app password (workstations), Restart pool.
Allow FirewallOn the new PC. Then Check now in Recover Slate.
Workstations followEach one finds the new address in server.json on the shared folder at its next start, if it has signed in to this server before.

A workstation follows server.json only when its saved address does not answer, the server named there accepts its password, and that server's id (kept in the database, so a restore keeps it) is the one it used before. Otherwise it asks first, and a person who is not sure answers No and uses Reconfigure server / database. Another PC answering on the network never takes the server's place. Data lost: at most the hour since the last backup.

A database kept on the shared folder

Older advice put the Database folder on the share. It still works after an upgrade, but the database can be damaged there and Recover Slate refuses to run. In Settings, press Move the database to this PC…: Slate backs up, stops the server, copies the database with slate_recovery and pgbouncer to %LOCALAPPDATA%\Slate_Central\LocalDatabase, checks the copy and starts from it. The old folder is not touched: delete it yourself after a week of clean running. If the server does not start from the copy, it goes back to the old folder. Typing a network path into Database folder asks first.

When something goes wrong

First, every timeOn the server PC: Operations → Run diagnostics and Recovery → Check now. On a workstation: Ctrl+Shift+D shows Slate's own diagnostics, including "Database problem at start-up".

What you see	What it means	What to do
Sign-in: "That user name and password do not match. Try again."	Wrong user name or password	Check both. Reset it in Users & Roles. If no administrator can sign in: Recover Slate → Reset password and switch the account back on.
Sign-in: "This studio's database is new and has no accounts yet."	Only the starter account exists	Make the administrator (see The first administrator), then create people.
Sign-in: "This account is switched off or its last day has passed."	`refuse_inactive_signin` is on and the account is deactivated or past its last day	Reactivate in Users & Roles, or Recover Slate → Reset password and switch the account back on.
Sign-in: "Your password is stored in an old, unsafe way..."	`no_plaintext_passwords` is on	Reset password for that person.
Button reads Database unavailable, with Try again	The workstation cannot reach the database	Server PC: is it Server Online? Firewall rules in place? The PC's address changed? Then Try again.
Database unavailable on one PC only, server fine	Often: this PC does not have the studio's database password	Ctrl+Shift+D or Settings says "This PC does not have the studio's database password." Recover Slate → Show app password; on the PC, Reconfigure server / database.
Many PCs cut off right after a password change	The server's password changed and they do not have it	Recover Slate → Show app password, then Set app password (workstations) with the previous password it lists. Restart pool.
"...Slate is working on this computer's copy." and LOCAL MODE in the header	Slate could not reach the server and opened this PC's local copy	Fix the connection. Changes are kept and sent with Sync afterwards. The Admin Panel does not work meanwhile.
Pop-up: Studio folder not found	The shared `Slate_Central` is not reachable from this PC	Map the drive or fix the network. Until then attendance, users and previews are not shared, Live Ops cannot see the PC, and updates do not arrive.
"...is not the file that was published (its checksum does not match)..."	The copied installer differs from what was published (a broken copy, or someone changed the file)	Publish it again from Slate Server. Never copy installers into `Updates\releases` by hand.
Publish update: "...is not a Slate workstation installer."	The file name is not `setup_Slate_Studio_v….exe` or `setup_Slate_Ops_v….exe`	Use the installers exactly as named in the release. Renamed files are refused.
Check for updates: "No update has been published for this Slate..."	Nothing in `Updates\releases`, or this PC looks at another shared folder	Publish, or check this PC's Server root.
Slate will not start after an update	A broken install	Run the installer from `Slate_Central\Updates\releases` on that PC by hand. Settings are kept.
Server: Tables in Database 0 on a studio that has data	The server is serving an empty folder	If the warning names "another database on this machine", point Settings → Database folder at it, Save, power off and on. Never create a new one.
Server: "...not the path in its settings..."	The configured folder was unreachable	Make the drive available, then power off and on.
Server: Connection Pool says wrong database or stopped	The pool does not match the settings	Restart pool.
Server: "The database could not be read: ..."	Often the server has no password to log in with	Recover Slate → Check now. "Server login" failing: Set superuser password (this server).
Nobody can sign in at all, or no administrator is left	Locked out	Recover Slate on the server PC with the Recovery Key: Create or restore administrator.
People stopped working after a switch was turned on	The switch refuses something they need	Recover Slate → choose it (or Every security switch) → Turn off. Works with the database down.
Recover Slate: "...is on a network drive" or "There is no Slate database in..."	Not the server PC, or the wrong folder	Run it on the server PC, signed in as the Windows account that runs Slate Server.
Recover Slate: "That is not this server's Recovery Key." / "Too many wrong keys."	Wrong key, or too many tries	Wait as told and try again. Lost it: Make a new Recovery Key as a Windows administrator.
One workstation slows everyone	A session is stuck holding the database	Analytics → select it → Disconnect selected.
Slate closes by itself, nothing in the log	A crash inside a picture or video library	Send `%LOCALAPPDATA%\Slate\Logs\native_crash.log` with the report.
"No free machines" when issuing	Nothing Available in Hardware	Add the machine in Hardware, or finish a repair.
Assign to: "Nobody holds the manage_it ability yet"	No role has Work the IT desk	Tick it on the IT role in Roles & Permissions.

Slate BETA 2.3.0 · Slate for IT · Keep a printed copy with the Recovery Key, away from the server PC. Each part starts on a new page.