f008

Documentation

Index

• Constants
• func F008CheckAutovacuumCost(report F008Report, a002Report a002.A002Report, result checkup.ReportResult) (checkup.ReportResult, error)
• func F008CheckMaxWorkers(report F008Report, a001Report a001.A001Report, result checkup.ReportResult) (checkup.ReportResult, error)
• func F008PreprocessReportData(data map[string]interface{})
• func F008Process(report F008Report, a001Report a001.A001Report, a002Report a002.A002Report) (checkup.ReportResult, error)
• type F008Report
• type F008ReportHostResult
• type F008ReportHostsResults
• type F008Setting

Constants

const AUTOVACUUM_MAX_WORKERS_DEFAULT int = 3

const AUTOVACUUM_VACUUM_COST_DELAY_DEFAULT int = 20

const AUTOVACUUM_VACUUM_COST_LIMIT_DEFAULT int = -1

const CPU_COUNT int = 16

const F008_DELAY_NOT_TUNED string = "F008_DELAY_NOT_TUNED"

const F008_DELAY_TUNE string = "F008_DELAY_TUNE"

const F008_MAX_WORKERS_LOW string = "F008_MAX_WORKERS_LOW"

const MSG_AUTOVACUUM_COST_DELAY_NOT_TUNED_CONCLUSION string = "[P1] Autovacuum cost delay and limit are not tuned. In Postgres versions prior to version 12, " +
	"the effective values of `autovacuum_vacuum_cost_limit` and `autovacuum_vacuum_cost_delay` are too " +
	"conservative, so autovacuum is overly throttled. Roughly speaking, the default settings mean, that all " +
	"autovacuum workers (except which currently processing tables with individual, per-table settings) can read data with combined read throughput only up to " +
	"~8 MiB/s. This is extremely low for modern disk systems, and with growing sizes of tables and indexes it " +
	"might lead to cases when some objects are processed by autovacuum during many hours: for example, it will take ~30 hours " +
	"to read 1 TiB of data if the allowed throughput is just 8 MiB/s. In some cases, it might lead to serious issues " +
	"such as performance degradation due to inability to process big tables in timed fashion and growing bloat, " +
	"and even to such critical issues as transaction ID wraparound. At the same time, if disk system is powerful " +
	"enough these risks can be easily mitigated by reducing throttling for autovacuum. In Postgres 12, it was " +
	"decided to decrease `autovacuum_vacuum_cost_delay` 10 times (from 20 ms to 2 ms)."

const MSG_AUTOVACUUM_COST_DELAY_NOT_TUNED_RECOMMENDATION string = "[P1] Autovacuum cost delay and limit are not tuned. Consider raising `autovacuum_vacuum_cost_limit` or " +
	"decreasing `autovacuum_vacuum_cost_delay`: for example, consider decreasing `autovacuum_vacuum_cost_delay` to 2 milliseconds, as it was done in default  " +
	"values in Postgres 12. For more fine-grained tuning, analyze disk capabilities (first of all, random read and random write troughput) and observe logs with " +
	"`log_autovacuum_min_duration = 0`, and perform several iterations of tuning."

const MSG_AUTOVACUUM_COST_DELAY_TUNE_RECOMMENDATION string = "Useful links related to autovacuum tuning:  \n" +
	"    - [PostgreSQL Documentation. 19.10. Automatic Vacuuming](https://www.postgresql.org/docs/%.1f/runtime-config-autovacuum.html)  \n" +
	"    - [Autovacuum Tuning Basics](https://www.2ndquadrant.com/en/blog/autovacuum-tuning-basics/) (2ndQuadrant, 2017)  \n" +
	"    - [Visualizing & Tuning Postgres Autovacuum](https://pganalyze.com/blog/visualizing-and-tuning-postgres-autovacuum) (pganalyze, 2017)  \n" +
	"    - [A Case Study of Tuning Autovacuum in Amazon RDS for PostgreSQL](https://aws.amazon.com/ru/blogs/database/a-case-study-of-tuning-autovacuum-in-amazon-rds-for-postgresql/) (AWS, 2018)  \n" +
	"    - [Understanding Autovacuum](https://www.youtube.com/watch?v=GqrBp0gyNHs) (video, 55 min, Citus Data, 2016)"

const MSG_MAX_WORKERS_LOW_CONCLUSION string = "[P1] Maximum number of autovacuum workers is too low. System has %d CPUs (see A001), but only 3 autovacuum " +
	"workers are allowed (according to the `autovacuum_max_workers` setting). This means that with increasing " +
	"workload, if significant part of this workload consists of modifying queries, autovacuum might start " +
	"lagging, maxing out `autovacuum_max_workers` value.\n"

const MSG_MAX_WORKERS_LOW_RECOMMENDATION string = "[P1] Maximum number of autovacuum workers is too low. Consider raising `autovacuum_max_workers`.\n" +
	"Consider values from %d to %d, depending on your workload. However, if 100%% or almost 100%% of the " +
	"workload is read-only queries and/or number of tables is less than 10, it might make sense to leave the " +
	"default value of `autovacuum_max_workers`. Refer to K002 to understand the workload and to L001 to see " +
	"the table list with sizes."

const VACUUM_COST_LIMIT_DEFAULT int = 200

Functions

func F008CheckAutovacuumCost

func F008CheckAutovacuumCost(report F008Report, a002Report a002.A002Report,
	result checkup.ReportResult) (checkup.ReportResult, error)

func F008CheckMaxWorkers

func F008CheckMaxWorkers(report F008Report, a001Report a001.A001Report,
	result checkup.ReportResult) (checkup.ReportResult, error)

func F008PreprocessReportData

func F008PreprocessReportData(data map[string]interface{})

func F008Process

func F008Process(report F008Report, a001Report a001.A001Report,
	a002Report a002.A002Report) (checkup.ReportResult, error)

Types

type F008Report

type F008Report struct {
	Project       string                  `json:"project"`
	Name          string                  `json:"name"`
	CheckId       string                  `json:"checkId"`
	Timestamptz   string                  `json:"timestamptz"`
	Database      string                  `json:"database"`
	Dependencies  map[string]interface{}  `json:"dependencies"`
	LastNodesJson checkup.ReportLastNodes `json:"last_nodes_json"`
	Results       F008ReportHostsResults  `json:"results"`
}

type F008ReportHostResult

type F008ReportHostResult struct {
	Data      map[string]F008Setting  `json:"data"`
	NodesJson checkup.ReportLastNodes `json:"nodes.json"`
}

type F008ReportHostsResults

type F008ReportHostsResults map[string]F008ReportHostResult

type F008Setting

type F008Setting struct {
	Name           string `json:"name"`
	Setting        string `json:"setting"`
	Unit           string `json:"unit"`
	Сategory       string `json:"category"`
	ShortDesc      string `json:"short_desc"`
	ExtraDesc      string `json:"extra_desc"`
	Context        string `json:"context"`
	Vartype        string `json:"vartype"`
	Source         string `json:"source"`
	MinVal         string `json:"min_val"`
	MaxVal         string `json:"max_val"`
	Numvals        string `json:"enumvals"`
	BootVal        string `json:"boot_val"`
	ResetVal       string `json:"reset_val"`
	Sourcefile     string `json:"sourcefile"`
	Sourceline     int    `json:"sourceline"`
	PendingRestart bool   `json:"pending_restart"`
}