Move the database to Postgres, before teams need the schema
First step of #1, and it goes first for one reason: #4 adds a team_id to nearly every table, and doing that twice -- once for SQLite, once for Postgres -- is work nobody gets paid for. The teams migrations now only have to be written against one database. The ten SQLite migrations are replaced by a single Postgres baseline rather than ported one by one. They were incremental in a way that has no value on a fresh install: 004 adds columns 008 drops again, and 008's backfill rewrites data a Postgres database never had. The history stays in git; the schema they add up to is now 001_baseline.sql. Timestamps stay BIGINT unix seconds and are NOT converted to timestamptz. Everything in Go already speaks epochs, so converting would have been a second, larger change riding along inside this one. It is worth doing on its own. The JSON columns did move to jsonb, because #4 will want to filter and index on labels. Most of the port is mechanical -- 170 placeholders from ? to $1 -- but four things needed more than a search and replace: * Dynamically built WHERE clauses cannot keep their numbering straight by hand, so they hand out placeholders through sqlArgs instead. A filter can now be added or reordered without renumbering anything. * SUM(resolved_at IS NULL) was SQLite counting a boolean as 0 or 1. Postgres has no sum(boolean), and this was breaking every dead man's switch -- silently, since the sweeper only logs. Now COUNT(*) FILTER. * unixepoch() became FLOOR(EXTRACT(EPOCH FROM now()))::bigint. The FLOOR is load-bearing: a bare cast rounds half up, so a row written at .6 of a second claimed a timestamp a second in the future and disagreed with the time.Now().Unix() the Go side stamps. * The unique-violation check matched SQLite's error text. It matches SQLSTATE 23505 now, so a renamed constraint cannot turn a 409 back into a 500. Tests need a real Postgres, because there is no in-memory Postgres the way there was an in-memory SQLite. Each test gets its own schema on a shared server -- cheaper than a database each, and still isolated. TERDUT_TEST_DSN says where it is; `make test-db` starts one locally and ci.yaml runs one as a service container. An unset DSN fails the suite rather than skipping it: a run that quietly tests nothing is worse than one that does not run. TestMigration_BackfillCarriesAckAndComments is deleted along with the migrations it replayed. What it protected -- an upgrade not losing acknowledgements and comments -- now belongs to scripts/sqlite-to-postgres.go, which is build-tagged so the SQLite driver stays out of the server binary. Both are meant to be deleted once this install has migrated. The chart loses the PVC, the data volume and the python backup sidecar, and requires database.dsnSecret.name: it provisions no database and cannot guess where the credentials live, so a render without it is meant to fail. Backups move to where Postgres actually runs. The other half of that -- the postgresql CR, the k8up pg_dump annotation and the network policy -- is a change to the wrapper chart in Ryuvia/charts and is not in here. Verified rather than assumed: the gate is green with -race against Postgres 17, govulncheck and gitleaks are clean, and the migration script was run end to end against a SQLite database built at the old schema and seeded in every table. Ids survive, so incidents keep their numbers and every foreign key still points where it did; the identity sequences are moved past the copied ids, and a webhook after the migration opened incident 12 rather than colliding at 1.
This commit is contained in:
Niklas Ye
@@ -0,0 +1,182 @@
|
||||
-- The Postgres baseline: the schema as it stood at the end of the SQLite line,
|
||||
-- in one file rather than ten.
|
||||
--
|
||||
-- The ten SQLite migrations are in git history up to the commit that introduced
|
||||
-- this one, and they replay against nothing here: their shape was incremental
|
||||
-- (columns added, then dropped again in 008) and 008's backfill rewrote data
|
||||
-- that a Postgres install never had. An existing SQLite database is carried over
|
||||
-- by scripts/sqlite-to-postgres.go, which copies rows into this schema.
|
||||
--
|
||||
-- Two conventions inherited deliberately:
|
||||
--
|
||||
-- * Timestamps are BIGINT unix seconds, not timestamptz. Everything in Go
|
||||
-- already speaks epochs, and converting was a second change riding along
|
||||
-- with the port. Worth revisiting on its own.
|
||||
--
|
||||
-- * Ids are GENERATED BY DEFAULT, not ALWAYS, so the migration script can
|
||||
-- insert rows with their original ids and keep every foreign key intact.
|
||||
-- setval at the end of the copy puts the sequences past them.
|
||||
|
||||
CREATE TABLE users (
|
||||
id BIGINT GENERATED BY DEFAULT AS IDENTITY PRIMARY KEY,
|
||||
username TEXT NOT NULL UNIQUE,
|
||||
email TEXT NOT NULL UNIQUE,
|
||||
created_at BIGINT NOT NULL DEFAULT FLOOR(EXTRACT(EPOCH FROM now()))::bigint,
|
||||
-- Where this user's notifications go. NULL means they get none; incidents
|
||||
-- assigned to them fall back to the configured fallback topic.
|
||||
ntfy_topic TEXT,
|
||||
-- NULL means the user has no password and can only use API keys.
|
||||
password_hash TEXT
|
||||
);
|
||||
|
||||
CREATE TABLE api_keys (
|
||||
id BIGINT GENERATED BY DEFAULT AS IDENTITY PRIMARY KEY,
|
||||
user_id BIGINT NOT NULL REFERENCES users(id) ON DELETE CASCADE,
|
||||
key_hash TEXT NOT NULL UNIQUE,
|
||||
name TEXT NOT NULL,
|
||||
created_at BIGINT NOT NULL DEFAULT FLOOR(EXTRACT(EPOCH FROM now()))::bigint,
|
||||
last_used_at BIGINT
|
||||
);
|
||||
|
||||
-- A session is a browser's credential, the cookie counterpart of an API key:
|
||||
-- only the hash of the token is stored. expires_at slides forward while the
|
||||
-- session is in use, so an on-call phone stays signed in.
|
||||
CREATE TABLE sessions (
|
||||
id BIGINT GENERATED BY DEFAULT AS IDENTITY PRIMARY KEY,
|
||||
token_hash TEXT NOT NULL UNIQUE,
|
||||
user_id BIGINT NOT NULL REFERENCES users(id) ON DELETE CASCADE,
|
||||
created_at BIGINT NOT NULL,
|
||||
last_seen_at BIGINT NOT NULL,
|
||||
expires_at BIGINT NOT NULL,
|
||||
user_agent TEXT
|
||||
);
|
||||
|
||||
CREATE INDEX idx_sessions_user ON sessions(user_id);
|
||||
|
||||
-- The machine-owned signal record: what Alertmanager says is true right now.
|
||||
-- Workflow state lives on incidents, never here, because the webhook upsert owns
|
||||
-- these rows and would overwrite it.
|
||||
CREATE TABLE alerts (
|
||||
id BIGINT GENERATED BY DEFAULT AS IDENTITY PRIMARY KEY,
|
||||
fingerprint TEXT NOT NULL UNIQUE,
|
||||
name TEXT NOT NULL,
|
||||
status TEXT NOT NULL CHECK (status IN ('firing', 'resolved')),
|
||||
labels JSONB NOT NULL DEFAULT '{}'::jsonb,
|
||||
annotations JSONB NOT NULL DEFAULT '{}'::jsonb,
|
||||
starts_at BIGINT NOT NULL,
|
||||
ends_at BIGINT,
|
||||
generator_url TEXT NOT NULL DEFAULT '',
|
||||
received_at BIGINT NOT NULL DEFAULT FLOOR(EXTRACT(EPOCH FROM now()))::bigint,
|
||||
archived_at BIGINT,
|
||||
-- Why the alert left the firing state: 'alertmanager' when a resolved
|
||||
-- webhook set it, 'expiry' when the sweeper inferred it from staleness.
|
||||
resolution_source TEXT
|
||||
);
|
||||
|
||||
CREATE INDEX alerts_status_idx ON alerts(status);
|
||||
CREATE INDEX alerts_name_idx ON alerts(name);
|
||||
CREATE INDEX alerts_received_at_idx ON alerts(received_at DESC);
|
||||
CREATE INDEX alerts_archived_at_idx ON alerts(archived_at);
|
||||
|
||||
CREATE TABLE schedule_entries (
|
||||
id BIGINT GENERATED BY DEFAULT AS IDENTITY PRIMARY KEY,
|
||||
user_id BIGINT NOT NULL REFERENCES users(id) ON DELETE CASCADE,
|
||||
date TEXT NOT NULL UNIQUE, -- YYYY-MM-DD; one person per day
|
||||
created_at BIGINT NOT NULL DEFAULT FLOOR(EXTRACT(EPOCH FROM now()))::bigint
|
||||
);
|
||||
|
||||
CREATE INDEX schedule_entries_date_idx ON schedule_entries(date);
|
||||
|
||||
-- The human work item: what people acknowledge, assign, snooze, discuss and
|
||||
-- resolve. Correlation uses Alertmanager's own groupKey, so incidents follow the
|
||||
-- group_by routing tree the operator already tuned.
|
||||
CREATE TABLE incidents (
|
||||
id BIGINT GENERATED BY DEFAULT AS IDENTITY PRIMARY KEY,
|
||||
group_key TEXT NOT NULL, -- Alertmanager groupKey, opaque
|
||||
title TEXT NOT NULL, -- rendered from group_labels
|
||||
group_labels JSONB NOT NULL DEFAULT '{}'::jsonb,
|
||||
status TEXT NOT NULL CHECK (status IN ('triggered', 'acknowledged', 'resolved')),
|
||||
severity TEXT, -- highest `severity` label across firing members
|
||||
triggered_at BIGINT NOT NULL,
|
||||
acknowledged_by BIGINT REFERENCES users(id) ON DELETE SET NULL,
|
||||
acknowledged_at BIGINT,
|
||||
assigned_to BIGINT REFERENCES users(id) ON DELETE SET NULL,
|
||||
snoozed_until BIGINT,
|
||||
resolved_at BIGINT,
|
||||
resolution_source TEXT, -- 'alerts' | 'manual'
|
||||
archived_at BIGINT
|
||||
);
|
||||
|
||||
-- Load-bearing: at most one OPEN incident per group_key. This is what makes
|
||||
-- "resolved incident + a new alert occurrence = a new incident" work, and it is
|
||||
-- the constraint the webhook's find-or-open lookup relies on.
|
||||
CREATE UNIQUE INDEX incidents_open_group_key_idx ON incidents(group_key) WHERE resolved_at IS NULL;
|
||||
CREATE INDEX incidents_status_idx ON incidents(status);
|
||||
CREATE INDEX incidents_triggered_at_idx ON incidents(triggered_at DESC);
|
||||
CREATE INDEX incidents_archived_at_idx ON incidents(archived_at);
|
||||
|
||||
-- Membership is historical, not a pointer on alerts: one alert row (one
|
||||
-- fingerprint) resolves and re-fires over time and belongs to a different
|
||||
-- incident each occurrence.
|
||||
CREATE TABLE incident_alerts (
|
||||
incident_id BIGINT NOT NULL REFERENCES incidents(id) ON DELETE CASCADE,
|
||||
alert_id BIGINT NOT NULL REFERENCES alerts(id) ON DELETE CASCADE,
|
||||
added_at BIGINT NOT NULL DEFAULT FLOOR(EXTRACT(EPOCH FROM now()))::bigint,
|
||||
PRIMARY KEY (incident_id, alert_id)
|
||||
);
|
||||
|
||||
CREATE INDEX incident_alerts_alert_id_idx ON incident_alerts(alert_id);
|
||||
|
||||
-- The timeline. Append-only, and the only history this server keeps: alert rows
|
||||
-- are mutated in place, so without this there is no record that anything
|
||||
-- happened. Notes are events too, so one query renders the whole story.
|
||||
CREATE TABLE incident_events (
|
||||
id BIGINT GENERATED BY DEFAULT AS IDENTITY PRIMARY KEY,
|
||||
incident_id BIGINT NOT NULL REFERENCES incidents(id) ON DELETE CASCADE,
|
||||
-- triggered | alert_added | alert_resolved | acknowledged | unacknowledged
|
||||
-- | assigned | snoozed | unsnoozed | resolved | note | notified | notify_failed
|
||||
type TEXT NOT NULL,
|
||||
user_id BIGINT REFERENCES users(id) ON DELETE SET NULL, -- NULL = the server acted
|
||||
alert_id BIGINT REFERENCES alerts(id) ON DELETE SET NULL,
|
||||
detail TEXT,
|
||||
created_at BIGINT NOT NULL DEFAULT FLOOR(EXTRACT(EPOCH FROM now()))::bigint
|
||||
);
|
||||
|
||||
CREATE INDEX incident_events_incident_idx ON incident_events(incident_id, created_at);
|
||||
|
||||
-- Delivery is an outbox rather than an inline HTTP call: a POST made while
|
||||
-- holding the webhook's transaction would hold a connection open across a
|
||||
-- network round trip. The webhook inserts a row; the notifier goroutine
|
||||
-- delivers it.
|
||||
CREATE TABLE notifications (
|
||||
id BIGINT GENERATED BY DEFAULT AS IDENTITY PRIMARY KEY,
|
||||
incident_id BIGINT NOT NULL REFERENCES incidents(id) ON DELETE CASCADE,
|
||||
-- Nullable: a notification sent to the fallback topic belongs to nobody,
|
||||
-- because nobody was on call when the incident opened.
|
||||
user_id BIGINT REFERENCES users(id) ON DELETE SET NULL,
|
||||
topic TEXT NOT NULL, -- resolved at enqueue: who was on call then
|
||||
kind TEXT NOT NULL CHECK (kind IN ('triggered', 'reminder', 'resolved')),
|
||||
created_at BIGINT NOT NULL,
|
||||
send_after BIGINT NOT NULL, -- retry backoff watermark
|
||||
attempts BIGINT NOT NULL DEFAULT 0,
|
||||
sent_at BIGINT,
|
||||
last_error TEXT -- kept after the last attempt, for debugging
|
||||
);
|
||||
|
||||
-- The delivery loop's only query: what is due and still unsent.
|
||||
CREATE INDEX notifications_pending_idx ON notifications(send_after) WHERE sent_at IS NULL;
|
||||
-- Reminders and resolved notices both look up an incident's newest row.
|
||||
CREATE INDEX notifications_incident_idx ON notifications(incident_id, id DESC);
|
||||
|
||||
-- A notification body is stored on the ntfy server and cached on the device, so
|
||||
-- a real API key must never appear in one. Each delivery mints its own token
|
||||
-- instead: one incident, one action, one day.
|
||||
CREATE TABLE incident_ack_tokens (
|
||||
token_hash TEXT PRIMARY KEY, -- SHA-256 of the raw token, as with api_keys
|
||||
incident_id BIGINT NOT NULL REFERENCES incidents(id) ON DELETE CASCADE,
|
||||
user_id BIGINT NOT NULL REFERENCES users(id) ON DELETE CASCADE,
|
||||
created_at BIGINT NOT NULL,
|
||||
expires_at BIGINT NOT NULL
|
||||
);
|
||||
|
||||
CREATE INDEX incident_ack_tokens_expires_idx ON incident_ack_tokens(expires_at);
|
||||
Reference in New Issue
Block a user