一、背景
对于PostgreSQL处理MVCC（数据文件中新增tuple）的方式；相比其他数据库(Oracle、Mysql)而言；更容易触发表/索引膨胀。因为update操作也会影响表膨胀的问题。PostgreSQL处理的方式是对表autovacuum，vacuum是不会降低水位线。能避免表、索引膨胀。vacuum full,reindex才会降低水位线。
当然通过update带来的表膨胀的情况还可以接受；PostgreSQL处理的方式是对表autovacuum。所以能autovacuum的一般不会对表膨胀带来大的影响。影响autovacuum进行对表回收的情况；大致有3种情况
长事务：数据库上有长时间没有提交的事务
SELECT
*
FROM
pg_stat_activity
WHERE
STATE = 'Idle in transaction';
未提交的2pc事务：这个特性是为分布式功能扩展的，非分布式的架构默认是关闭的
SELECT
gid,
PREPARED,
OWNER,
DATABASE,
TRANSACTION AS xmin
FROM
pg_prepared_xacts
ORDER BY
age( TRANSACTION ) DESC
复制槽：逻辑复制、流复制都会有复制槽；复制槽的作用：通过记录复制的当前位置；保证备库、订阅端未接收到的数据不会在主库删除。但是废弃的复制槽会影响表的vacuum。这个是危害最大的
SELECT
slot_name,
slot_type,
DATABASE,
xmin
FROM
pg_replication_slots
ORDER BY
age( xmin ) DESC;
二、监控
如何知道表膨胀呢？即监控
2.1、pgstattuple插件
# 安装
create extension pgstattuple;
# 查看前5的膨胀表
select oid::regclass,(pgstattuple(oid)).* from pg_class where relkind='r' order by free_space desc limit 5 offset 0;
# 查看前5的索引
select oid::regclass,(pgstattuple(oid)).* from pg_class where relkind='i' order by free_space desc limit 5 offset 0;
2.2 bucardo发布postgres_check中包含的查询膨胀的SQL
-- 表
SELECT
current_database() AS db, schemaname, tablename, reltuples::bigint AS tups, relpages::bigint AS pages, otta,
ROUND(CASE WHEN otta=0 OR sml.relpages=0 OR sml.relpages=otta THEN 0.0 ELSE sml.relpages/otta::numeric END,1) AS tbloat,
CASE WHEN relpages < otta THEN 0 ELSE relpages::bigint - otta END AS wastedpages,
CASE WHEN relpages < otta THEN 0 ELSE bs*(sml.relpages-otta)::bigint END AS wastedbytes,
CASE WHEN relpages < otta THEN $$0 bytes$$::text ELSE (bs*(relpages-otta))::bigint || $$ bytes$$ END AS wastedsize,
iname, ituples::bigint AS itups, ipages::bigint AS ipages, iotta,
ROUND(CASE WHEN iotta=0 OR ipages=0 OR ipages=iotta THEN 0.0 ELSE ipages/iotta::numeric END,1) AS ibloat,
CASE WHEN ipages < iotta THEN 0 ELSE ipages::bigint - iotta END AS wastedipages,
CASE WHEN ipages < iotta THEN 0 ELSE bs*(ipages-iotta) END AS wastedibytes,
CASE WHEN ipages < iotta THEN $$0 bytes$$ ELSE (bs*(ipages-iotta))::bigint || $$ bytes$$ END AS wastedisize,
CASE WHEN relpages < otta THEN
CASE WHEN ipages < iotta THEN 0 ELSE bs*(ipages-iotta::bigint) END
ELSE CASE WHEN ipages < iotta THEN bs*(relpages-otta::bigint)
ELSE bs*(relpages-otta::bigint + ipages-iotta::bigint) END
END AS totalwastedbytes
FROM (
SELECT
nn.nspname AS schemaname,
cc.relname AS tablename,
COALESCE(cc.reltuples,0) AS reltuples,
COALESCE(cc.relpages,0) AS relpages,
COALESCE(bs,0) AS bs,
COALESCE(CEIL((cc.reltuples*((datahdr+ma-
(CASE WHEN datahdr%ma=0 THEN ma ELSE datahdr%ma END))+nullhdr2+4))/(bs-20::float)),0) AS otta,
COALESCE(c2.relname,$$?$$) AS iname, COALESCE(c2.reltuples,0) AS ituples, COALESCE(c2.relpages,0) AS ipages,
COALESCE(CEIL((c2.reltuples*(datahdr-12))/(bs-20::float)),0) AS iotta -- very rough approximation, assumes all cols
FROM
pg_class cc
JOIN pg_namespace nn ON cc.relnamespace = nn.oid AND nn.nspname <> $$information_schema$$
LEFT JOIN
(
SELECT
ma,bs,foo.nspname,foo.relname,
(datawidth+(hdr+ma-(case when hdr%ma=0 THEN ma ELSE hdr%ma END)))::numeric AS datahdr,
(maxfracsum*(nullhdr+ma-(case when nullhdr%ma=0 THEN ma ELSE nullhdr%ma END))) AS nullhdr2
FROM (
SELECT
ns.nspname, tbl.relname, hdr, ma, bs,
SUM((1-coalesce(null_frac,0))*coalesce(avg_width, 2048)) AS datawidth,
MAX(coalesce(null_frac,0)) AS maxfracsum,
hdr+(
SELECT 1+count(*)/8
FROM pg_stats s2
WHERE null_frac<>0 AND s2.schemaname = ns.nspname AND s2.tablename = tbl.relname
) AS nullhdr
FROM pg_attribute att
JOIN pg_class tbl ON att.attrelid = tbl.oid
JOIN pg_namespace ns ON ns.oid = tbl.relnamespace
LEFT JOIN pg_stats s ON s.schemaname=ns.nspname
AND s.tablename = tbl.relname
AND s.inherited=false
AND s.attname=att.attname,
(
SELECT
(SELECT current_setting($$block_size$$)::numeric) AS bs,
CASE WHEN SUBSTRING(SPLIT_PART(v, $$ $$, 2) FROM $$#"[0-9]+.[0-9]+#"%$$ for $$#$$)
IN ($$8.0$$,$$8.1$$,$$8.2$$) THEN 27 ELSE 23 END AS hdr,
CASE WHEN v ~ $$mingw32$$ OR v ~ $$64-bit$$ THEN 8 ELSE 4 END AS ma
FROM (SELECT version() AS v) AS foo
) AS constants
WHERE att.attnum > 0 AND tbl.relkind=$$r$$
GROUP BY 1,2,3,4,5
) AS foo
) AS rs
ON cc.relname = rs.relname AND nn.nspname = rs.nspname
LEFT JOIN pg_index i ON indrelid = cc.oid
LEFT JOIN pg_class c2 ON c2.oid = i.indexrelid
) AS sml order by wastedbytes desc limit 5
-- 索引
SELECT
current_database() AS db, schemaname, tablename, reltuples::bigint AS tups, relpages::bigint AS pages, otta,
ROUND(CASE WHEN otta=0 OR sml.relpages=0 OR sml.relpages=otta THEN 0.0 ELSE sml.relpages/otta::numeric END,1) AS tbloat,
CASE WHEN relpages < otta THEN 0 ELSE relpages::bigint - otta END AS wastedpages,
CASE WHEN relpages < otta THEN 0 ELSE bs*(sml.relpages-otta)::bigint END AS wastedbytes,
CASE WHEN relpages < otta THEN $$0 bytes$$::text ELSE (bs*(relpages-otta))::bigint || $$ bytes$$ END AS wastedsize,
iname, ituples::bigint AS itups, ipages::bigint AS ipages, iotta,
ROUND(CASE WHEN iotta=0 OR ipages=0 OR ipages=iotta THEN 0.0 ELSE ipages/iotta::numeric END,1) AS ibloat,
CASE WHEN ipages < iotta THEN 0 ELSE ipages::bigint - iotta END AS wastedipages,
CASE WHEN ipages < iotta THEN 0 ELSE bs*(ipages-iotta) END AS wastedibytes,
CASE WHEN ipages < iotta THEN $$0 bytes$$ ELSE (bs*(ipages-iotta))::bigint || $$ bytes$$ END AS wastedisize,
CASE WHEN relpages < otta THEN
CASE WHEN ipages < iotta THEN 0 ELSE bs*(ipages-iotta::bigint) END
ELSE CASE WHEN ipages < iotta THEN bs*(relpages-otta::bigint)
ELSE bs*(relpages-otta::bigint + ipages-iotta::bigint) END
END AS totalwastedbytes
FROM (
SELECT
nn.nspname AS schemaname,
cc.relname AS tablename,
COALESCE(cc.reltuples,0) AS reltuples,
COALESCE(cc.relpages,0) AS relpages,
COALESCE(bs,0) AS bs,
COALESCE(CEIL((cc.reltuples*((datahdr+ma-
(CASE WHEN datahdr%ma=0 THEN ma ELSE datahdr%ma END))+nullhdr2+4))/(bs-20::float)),0) AS otta,
COALESCE(c2.relname,$$?$$) AS iname, COALESCE(c2.reltuples,0) AS ituples, COALESCE(c2.relpages,0) AS ipages,
COALESCE(CEIL((c2.reltuples*(datahdr-12))/(bs-20::float)),0) AS iotta -- very rough approximation, assumes all cols
FROM
pg_class cc
JOIN pg_namespace nn ON cc.relnamespace = nn.oid AND nn.nspname <> $$information_schema$$
LEFT JOIN
(
SELECT
ma,bs,foo.nspname,foo.relname,
(datawidth+(hdr+ma-(case when hdr%ma=0 THEN ma ELSE hdr%ma END)))::numeric AS datahdr,
(maxfracsum*(nullhdr+ma-(case when nullhdr%ma=0 THEN ma ELSE nullhdr%ma END))) AS nullhdr2
FROM (
SELECT
ns.nspname, tbl.relname, hdr, ma, bs,
SUM((1-coalesce(null_frac,0))*coalesce(avg_width, 2048)) AS datawidth,
MAX(coalesce(null_frac,0)) AS maxfracsum,
hdr+(
SELECT 1+count(*)/8
FROM pg_stats s2
WHERE null_frac<>0 AND s2.schemaname = ns.nspname AND s2.tablename = tbl.relname
) AS nullhdr
FROM pg_attribute att
JOIN pg_class tbl ON att.attrelid = tbl.oid
JOIN pg_namespace ns ON ns.oid = tbl.relnamespace
LEFT JOIN pg_stats s ON s.schemaname=ns.nspname
AND s.tablename = tbl.relname
AND s.inherited=false
AND s.attname=att.attname,
(
SELECT
(SELECT current_setting($$block_size$$)::numeric) AS bs,
CASE WHEN SUBSTRING(SPLIT_PART(v, $$ $$, 2) FROM $$#"[0-9]+.[0-9]+#"%$$ for $$#$$)
IN ($$8.0$$,$$8.1$$,$$8.2$$) THEN 27 ELSE 23 END AS hdr,
CASE WHEN v ~ $$mingw32$$ OR v ~ $$64-bit$$ THEN 8 ELSE 4 END AS ma
FROM (SELECT version() AS v) AS foo
) AS constants
WHERE att.attnum > 0 AND tbl.relkind=$$r$$
GROUP BY 1,2,3,4,5
) AS foo
) AS rs
ON cc.relname = rs.relname AND nn.nspname = rs.nspname
LEFT JOIN pg_index i ON indrelid = cc.oid
LEFT JOIN pg_class c2 ON c2.oid = i.indexrelid
) AS sml order by wastedibytes desc limit 5
三、处理表/索引膨胀
使用pg_reorg|pg_repack、pg_squeeze或者vacuum full可以回收膨胀的空间
--http://www.postgres.cn/news/viewone/1/224
索引膨胀会影响查询效率；处理索引膨胀的方法：重建索引
创建新索引
create index CONCURRENTLY new_index
删除旧索引
drop index new_index
或者
reindex index ... CONCURRENTLY
四、避免表/索引膨胀
4.1、实例级调整
合理调整autovacuum参数;
4.2、数据表级调整
1、 设置合适的autovacuum_vacuum_scale_factor, 大表如果频繁的有更新或删除和插入操作, 建议设置较小的autovacuum_vacuum_scale_factor来降低浪费空间，加快对表的vacuum操作频率
对更新频繁的表，单独调整
alter table tablename set (autovacuum_vacuum_scale_factor=0.05);
2、设置表的fillfactor；对频繁更新的表；调低fillfactor参数
alter table tablename set (fillfactor = 85)