on this page

Switching toolsMoving from Squadcast

Moving from Squadcast

Squadcast, sold today as SolarWinds Incident Response, keeps your on-call setup in schedules, escalation policies and services. Tissue Mast's importer reads all three through Squadcast's API and builds the same setup here: rotations, escalation policies and alert sources. This page covers what it takes, what it leaves behind, and how to move the senders afterwards.


Before you start

You need two things from Squadcast:

  • The API refresh token. Squadcast makes one on your profile page, under API token. It reads what its owner can see, so make it as someone who can see every team you want to bring over, usually an account owner or admin. Mast trades it for a short-lived access token for the length of one read, and keeps no copy of either, not in its records, its logs or the account's activity. Revoke the token in Squadcast when you're done.
  • The region. US accounts sign in at app.squadcast.com and EU accounts at app.eu.squadcast.com. A token works only in its own region.

On the Mast side, you need to be the account's owner or an admin. Inviting your team first saves a step later, since the importer matches people by email address. Each person installs the app and links a device, and someone with no linked device can't be on a rotation.

The importer

The importer is in the web app under On call, as Import from Squadcast. Paste the refresh token, pick the region and press Read Squadcast.

The first read is a dry run and changes nothing. It's open on every plan, Free included, so you can see what a move involves before you pick one. It lists every rotation, escalation policy and alert source Mast would make, and for each one whether it's new, changes something already there, is already the same, is left out, or doesn't fit the plan. A change lists each field with its old and new value. Anything that doesn't fit the plan stays on the list, marked over plan, and a trial of Team or Company covers rotations and policies.

Above the lists it sums up the move: how many Squadcast users there are and how many are already members here, how many teams the token can see and what belongs to each, how many people Squadcast texts or calls for a new incident, and the smallest Mast plan that holds the lot, with its monthly price. Under the lists, Left behind in this account names each setting it read and has nowhere to put, one line each, so you can decide what to rebuild by hand.

Untick anything you don't want, then press Import the items ticked. Mast reads Squadcast again at that moment, so a change made there in between is picked up. Everything is matched by name, so running it again changes what's there instead of making copies, and a run that failed partway finishes on the next one.

People

The importer never invites anyone on its own. It lists everyone in Squadcast and matches each one to a member of this account by email address, ignoring case. Everyone not on the account gets an unticked box, and Invite sends each person ticked a normal member invite, as a developer or an admin. On Free, invites count against the plan's 5 people, and an account can send at most 20 an hour.

Someone who isn't on the account yet is left out of their rotations and escalation steps, and the dry run names them on each one. Once they've joined, run the import again and it puts them back.

How each part maps

  • Teams. When the token sees more than one team, each schedule, policy and service gets its team's name in front: the Platform team's "Database" schedule becomes Platform / Database. With a single team, names stay as they are. Alert source names are lowercase and can't hold a slash, so the Platform team's "API" service becomes the alert source platform - api.
  • Schedules. Each rotation in a schedule becomes its own Mast rotation, named <schedule> / <rotation>, so a schedule with day and night rotations becomes two. Daily stays daily, weekly hands off on the same day and time in the schedule's zone, and every two weeks stays every two weeks. A rotation that changes hands after a number of hours or days becomes a custom length in hours, up to 720. Monthly rotations are left out with a note. A rotation that only covers part of the day or week, such as weeknights, is on around the clock in Mast, and the note says when it was on call in Squadcast. Paused schedules aren't imported.
  • Participants. A turn with one person carries over as it is. A turn shared by several people keeps the first and names the others, since a Mast turn is one person. A squad or a team in a rotation is left out with a note.
  • Overrides. Overrides over the next 90 days come across as timed overrides in the schedule's zone. An override handed to a squad or a team is left out, since a Mast override names one person. If Squadcast's API won't list overrides, the schedules still come across and the dry run says to add upcoming overrides by hand.
  • Escalation policies. Each level becomes a step after the same delay. A level that pages a schedule becomes a step for that rotation, and one that pages a user pages that person. Squads, teams and Squadcast's older schedules get the closest Mast step, and the dry run says which. Policy repeats carry over as going round again, up to 10 times. Round robin and a level's own repeats don't, and the dry run lists every policy that used them. A level that also texted or called is flagged, since Mast pages the phone app and calls only from a step set to call.
  • Services. Each service becomes an alert source attached to the escalation policy it names. Its tags, dependencies, Slack channel, alert grouping, auto-pause and delayed notifications stay in Squadcast, and the dry run lists any a service used. A service in a maintenance window in Squadcast pages from the start in Mast.

What doesn't carry over

The importer leaves these in Squadcast and lists them once: personal notification rules, on-call reminders, routing, tagging, suppression and deduplication rules, runbooks, status pages, SLOs, webhooks and extensions, incidents and incident history.

  • Personal notification rules. A person's own "SMS after 5 minutes, then call" has no equivalent. Mast pages the phone through push, and a policy step with Call ticked phones everyone it reaches who has confirmed a number (phone calls). There's no SMS or email paging.
  • Routing and deduplication rules. A Mast alert source pages its own policy. To split alerts by content, give each kind its own alert source. Repeats of an open page fold into it when the sender reuses the same dedupe key.
  • Status pages. Mast has its own status pages, built from your probes and vitals, so set them up there.
  • Incident history. Past incidents, notes and postmortems stay in Squadcast. Export what you want to keep before you cancel. Nothing on the Tissue Mast side reads them.

Moving the senders

Mast doesn't answer Squadcast's incident API, so every sender changes address. When the import finishes it prints the re-key list: every alert source with the service it came from and its send URL. The URL holds the alert source's key, so treat the list like a password. Channel URLs are readable back from GET /v1/mast/state at any time.

  • Anything with a generic webhook or a shell step posts to the alert source's send URL with Mast's own fields: title, body, key and priority. Send the same key again with resolve=1 and the page closes.
  • Alertmanager, Grafana and Zabbix each have a setup in the Opsgenie guide, since Mast answers Opsgenie's alert API. If one of them already has an Opsgenie integration, that's often the shortest way across.
  • Anything that sent email to a service's Squadcast address needs a real change, since Mast has no email ingest. A curl in the same cron line is the usual replacement.

Work through the list in this order:

  1. Move the senders first, while Squadcast still takes alerts, and send to both for a few days so you can compare.
  2. Send one test page per channel from the sender itself, and watch it land on the device that should get it.
  3. Check the rotations in the app for the coming weeks, against who Squadcast has on call. A person without a linked device shows up there as a gap.
  4. Turn off the Squadcast services once every sender has moved, then revoke the refresh token.

Terraform

The import also gives you a Terraform import block for everything it made or changed, under the re-key list, with Copy and Download buttons. Put the blocks in a .tf file and run terraform plan -generate-config-out=mast.tf, and Terraform writes the matching resources for the tissue provider. The Opsgenie guide shows what the blocks look like.

From the API

The importer's routes are POST /v1/mast/import/squadcast/plan and /apply, and they take an account token with the mast:write scope. Send {"token": "…", "region": "us"} to plan. On apply, an optional "select" holds lists of Mast names from the plan under schedules, policies and channels, and a kind left out imports everything of it that fits. POST /v1/mast/import/squadcast/invite takes up to 100 addresses to invite, and with an API token it needs the account:write scope. A ribo mast import squadcast command is coming soon.