Unify check attempt data type to uint32 already used somewhere

A float isn't necessary as in Icinga 2 Checkable#max_check_attempts and
check_attempt are ints. But uint8 isn't enough for e.g. 1 check/s to get
HARD after 5m (300s > 255).

cmd/icingadb-migrate/convert.go

@@ -724,8 +724,8 @@ type stateRow = struct {
 	StateTimeUsec       uint32
 	State               uint8
 	StateType           uint8
-	CurrentCheckAttempt uint16
-	MaxCheckAttempts    uint16
+	CurrentCheckAttempt uint32
+	MaxCheckAttempts    uint32
 	LastState           uint8
 	LastHardState       uint8
 	Output              sql.NullString
@@ -798,10 +798,10 @@ func convertStateRows(
 			HardState:         row.LastHardState,
 			PreviousSoftState: row.LastState,
 			PreviousHardState: previousHardState,
-			CheckAttempt:      uint8(row.CurrentCheckAttempt),
+			CheckAttempt:      row.CurrentCheckAttempt,
 			Output:            icingadbTypes.String{NullString: row.Output},
 			LongOutput:        icingadbTypes.String{NullString: row.LongOutput},
-			MaxCheckAttempts:  uint32(row.MaxCheckAttempts),
+			MaxCheckAttempts:  row.MaxCheckAttempts,
 			CheckSource:       icingadbTypes.String{NullString: row.CheckSource},
 		})
 

doc/04-Upgrading.md

@@ -3,6 +3,35 @@
 Specific version upgrades are described below. Please note that version upgrades are incremental.
 If you are upgrading across multiple versions, make sure to follow the steps for each of them.
 
+## Upgrading to Icinga DB v1.1.2
+
+Please apply the `1.1.2.sql` upgrade script to your database.
+For package installations, you can find this file at `/usr/share/icingadb/schema/mysql/upgrades/` or
+`/usr/share/icingadb/schema/pgsql/upgrades/`, depending on your database type.
+
+As the daemon checks the schema version, the recommended way to perform the upgrade is to stop the daemon, apply the
+schema upgrade and then start the new daemon version. If you want to minimize downtime as much as possible, it is safe
+to apply this schema upgrade while the Icinga DB v1.1.1 daemon is still running and then restart the daemon with the
+new version. Please keep in mind that depending on the distribution, your package manager may automatically attempt to
+restart the daemon when upgrading the package.
+
+### Upgrading the state_history Table (optional)
+
+Icinga DB crashes if hosts/services reach check attempt 256. The `check_attempt` column of the `state_history` table
+is too small to fit values greater 255. ([#655](https://github.com/Icinga/icingadb/issues/655))
+Databases created with the v1.1.2 schema are not affected, but upgraded ones are.
+
+The schema upgrade required to fix this isn't included in `1.1.2.sql` as it re-writes the whole `state_history` table.
+This can take a lot of time depending on the history size and the performance of the database.
+During this time that table will be locked exclusively and can't be accessed otherwise.
+This means that the existing history can't be viewed in Icinga Web and new history entries will be buffered in Redis.
+
+That optional table re-write is performed by the upgrade script `optional/1.1.2-history.sql`.
+It should be applied during the next longer maintenance window after `1.1.2.sql`.
+Do not interrupt it! At best use tmux/screen not to lose your SSH session.
+Until this upgrade you'll have to lower `max_check_attempts`. Optionally increase the respective `retry_interval`
+not to change the timespan from an outage to a hard state (`max_check_attempts * retry_interval`).
+
 ## Upgrading to Icinga DB v1.1.1
 
 Please apply the `1.1.1.sql` upgrade script to your database.

pkg/icingadb/db.go

@@ -85,8 +85,8 @@ func NewDb(db *sqlx.DB, logger *logging.Logger, options *Options) *DB {
 }
 
 const (
-	expectedMysqlSchemaVersion    = 4
-	expectedPostgresSchemaVersion = 2
+	expectedMysqlSchemaVersion    = 5
+	expectedPostgresSchemaVersion = 3
 )
 
 // CheckSchema asserts the database schema of the expected version being present.

pkg/icingadb/v1/checkable.go

@@ -30,7 +30,7 @@ type Checkable struct {
 	IconImageAlt          string       `json:"icon_image_alt"`
 	IconImageId           types.Binary `json:"icon_image_id"`
 	IsVolatile            types.Bool   `json:"is_volatile"`
-	MaxCheckAttempts      float64      `json:"max_check_attempts"`
+	MaxCheckAttempts      uint32       `json:"max_check_attempts"`
 	Notes                 string       `json:"notes"`
 	NotesUrlId            types.Binary `json:"notes_url_id"`
 	NotificationsEnabled  types.Bool   `json:"notifications_enabled"`

pkg/icingadb/v1/history/state.go

@@ -14,7 +14,7 @@ type StateHistory struct {
 	HardState          uint8           `json:"hard_state"`
 	PreviousSoftState  uint8           `json:"previous_soft_state"`
 	PreviousHardState  uint8           `json:"previous_hard_state"`
-	CheckAttempt       uint8           `json:"check_attempt"`
+	CheckAttempt       uint32          `json:"check_attempt"`
 	Output             types.String    `json:"output"`
 	LongOutput         types.String    `json:"long_output"`
 	MaxCheckAttempts   uint32          `json:"max_check_attempts"`

pkg/icingadb/v1/state.go

@@ -9,7 +9,7 @@ type State struct {
 	EnvironmentMeta           `json:",inline"`
 	AcknowledgementCommentId  types.Binary               `json:"acknowledgement_comment_id"`
 	LastCommentId             types.Binary               `json:"last_comment_id"`
-	CheckAttempt              uint8                      `json:"check_attempt"`
+	CheckAttempt              uint32                     `json:"check_attempt"`
 	CheckCommandline          types.String               `json:"check_commandline"`
 	CheckSource               types.String               `json:"check_source"`
 	SchedulingSource          types.String               `json:"scheduling_source"`

schema/mysql/schema.sql

@@ -292,7 +292,7 @@ CREATE TABLE host_state (
   hard_state tinyint unsigned NOT NULL,
   previous_soft_state tinyint unsigned NOT NULL,
   previous_hard_state tinyint unsigned NOT NULL,
-  check_attempt tinyint unsigned NOT NULL,
+  check_attempt int unsigned NOT NULL,
   severity smallint unsigned NOT NULL,
 
   output longtext DEFAULT NULL,
@@ -460,7 +460,7 @@ CREATE TABLE service_state (
   hard_state tinyint unsigned NOT NULL,
   previous_soft_state tinyint unsigned NOT NULL,
   previous_hard_state tinyint unsigned NOT NULL,
-  check_attempt tinyint unsigned NOT NULL,
+  check_attempt int unsigned NOT NULL,
   severity smallint unsigned NOT NULL,
 
   output longtext DEFAULT NULL,
@@ -1147,7 +1147,7 @@ CREATE TABLE state_history (
   hard_state tinyint unsigned NOT NULL,
   previous_soft_state tinyint unsigned NOT NULL,
   previous_hard_state tinyint unsigned NOT NULL,
-  check_attempt tinyint unsigned NOT NULL,
+  check_attempt int unsigned NOT NULL, -- may be a tinyint unsigned, see https://icinga.com/docs/icinga-db/latest/doc/04-Upgrading/#upgrading-to-icinga-db-v112
   output longtext DEFAULT NULL,
   long_output longtext DEFAULT NULL,
   max_check_attempts int unsigned NOT NULL,
@@ -1343,4 +1343,4 @@ CREATE TABLE icingadb_schema (
 ) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_bin ROW_FORMAT=DYNAMIC;
 
 INSERT INTO icingadb_schema (version, timestamp)
-  VALUES (4, UNIX_TIMESTAMP() * 1000);
+  VALUES (5, UNIX_TIMESTAMP() * 1000);

schema/mysql/upgrades/1.1.2.sql

@@ -1 +1,10 @@
+ALTER TABLE host_state MODIFY COLUMN check_attempt int unsigned NOT NULL;
+
+ALTER TABLE service_state MODIFY COLUMN check_attempt int unsigned NOT NULL;
+
+ALTER TABLE state_history MODIFY COLUMN check_attempt tinyint unsigned NOT NULL COMMENT 'optional schema upgrade not applied yet, see https://icinga.com/docs/icinga-db/latest/doc/04-Upgrading/#upgrading-to-icinga-db-v112';
+
 UPDATE icingadb_schema SET timestamp = UNIX_TIMESTAMP(timestamp / 1000) * 1000 WHERE timestamp > 20000000000000000;
+
+INSERT INTO icingadb_schema (version, timestamp)
+  VALUES (5, UNIX_TIMESTAMP() * 1000);

schema/mysql/upgrades/optional/1.1.2-history.sql

@@ -0,0 +1 @@
+ALTER TABLE state_history MODIFY COLUMN check_attempt int unsigned NOT NULL;

schema/pgsql/schema.sql

@@ -405,7 +405,7 @@ CREATE TABLE host_state (
   hard_state tinyuint NOT NULL,
   previous_soft_state tinyuint NOT NULL,
   previous_hard_state tinyuint NOT NULL,
-  check_attempt tinyuint NOT NULL,
+  check_attempt uint NOT NULL,
   severity smalluint NOT NULL,
 
   output text DEFAULT NULL,
@@ -675,7 +675,7 @@ CREATE TABLE service_state (
   hard_state tinyuint NOT NULL,
   previous_soft_state tinyuint NOT NULL,
   previous_hard_state tinyuint NOT NULL,
-  check_attempt tinyuint NOT NULL,
+  check_attempt uint NOT NULL,
   severity smalluint NOT NULL,
 
   output text DEFAULT NULL,
@@ -1846,7 +1846,7 @@ CREATE TABLE state_history (
   hard_state tinyuint NOT NULL,
   previous_soft_state tinyuint NOT NULL,
   previous_hard_state tinyuint NOT NULL,
-  check_attempt tinyuint NOT NULL,
+  check_attempt uint NOT NULL, -- may be a tinyuint, see https://icinga.com/docs/icinga-db/latest/doc/04-Upgrading/#upgrading-to-icinga-db-v112
   output text DEFAULT NULL,
   long_output text DEFAULT NULL,
   max_check_attempts uint NOT NULL,
@@ -2181,4 +2181,4 @@ CREATE TABLE icingadb_schema (
 ALTER SEQUENCE icingadb_schema_id_seq OWNED BY icingadb_schema.id;
 
 INSERT INTO icingadb_schema (version, timestamp)
-  VALUES (2, extract(epoch from now()) * 1000);
+  VALUES (3, extract(epoch from now()) * 1000);

schema/pgsql/upgrades/1.1.2.sql

@@ -137,3 +137,12 @@ BEGIN
   RETURN (100 * (total_time - problem_time)::decimal / total_time)::decimal(7, 4);
 END;
 $$;
+
+ALTER TABLE host_state ALTER COLUMN check_attempt TYPE uint;
+
+ALTER TABLE service_state ALTER COLUMN check_attempt TYPE uint;
+
+COMMENT ON COLUMN state_history.check_attempt IS 'optional schema upgrade not applied yet, see https://icinga.com/docs/icinga-db/latest/doc/04-Upgrading/#upgrading-to-icinga-db-v112';
+
+INSERT INTO icingadb_schema (version, timestamp)
+  VALUES (3, extract(epoch from now()) * 1000);

schema/pgsql/upgrades/optional/1.1.2-history.sql

@@ -0,0 +1,3 @@
+ALTER TABLE state_history ALTER COLUMN check_attempt TYPE uint;
+
+COMMENT ON COLUMN state_history.check_attempt IS NULL;