Merge in Pull Request #20 from Steeve Lennmark.

Adds support for repacking only the tables in a specified schema. This
doesn't support --only-indexes mode, but that seems alright for now.

Fix merge conflicts, and make a few tweaks along the way:
 * bump version to 1.3-dev0
 * add Beena to list of maintainers
 * documentation wordsmithing
 * fix up the INFO message printed for each index in --index or
   --only-indexes mode, so that it is only printed once per index, and
   prints the name of the original index, not that of the transient
   index_%u name.
This commit is contained in:
Josh Kupershmidt 2014-05-24 15:42:24 -04:00
commit f0cd787bf5
7 changed files with 105 additions and 13 deletions

View File

@ -2,18 +2,19 @@
"name": "pg_repack",
"abstract": "PostgreSQL module for data reorganization",
"description": "Reorganize tables in PostgreSQL databases with minimal locks",
"version": "1.2.1",
"version": "1.3-dev0",
"maintainer": [
"Josh Kupershmidt <schmiddy@gmail.com>",
"Daniele Varrazzo <daniele.varrazzo@gmail.com>"
"Daniele Varrazzo <daniele.varrazzo@gmail.com>",
"Beena Emerson <memissemerson@gmail.com>"
],
"tags": [ "bloat", "maintenance", "vacuum", "cluster" ],
"release_status": "stable",
"release_status": "testing",
"license": "bsd",
"provides": {
"pg_repack": {
"file": "lib/pg_repack.sql",
"version": "1.2.1",
"version": "1.3-dev0",
"abstract": "Reorganize tables in PostgreSQL databases with minimal locks"
}
},

View File

@ -219,6 +219,7 @@ static bool analyze = true;
static bool alldb = false;
static bool noorder = false;
static SimpleStringList table_list = {NULL, NULL};
static SimpleStringList schema_list = {NULL, NULL};
static char *orderby = NULL;
static char *tablespace = NULL;
static bool moveidx = false;
@ -240,6 +241,7 @@ static pgut_option options[] =
{
{ 'b', 'a', "all", &alldb },
{ 'l', 't', "table", &table_list },
{ 'l', 'c', "schema", &schema_list },
{ 'b', 'n', "no-order", &noorder },
{ 'b', 'N', "dry-run", &dryrun },
{ 's', 'o', "order-by", &orderby },
@ -308,6 +310,11 @@ main(int argc, char *argv[])
}
else
{
if (schema_list.head && table_list.head)
ereport(ERROR,
(errcode(EINVAL),
errmsg("cannot repack specific table(s) in schema, use schema.table notation instead")));
if (noorder)
orderby = "";
@ -317,6 +324,10 @@ main(int argc, char *argv[])
ereport(ERROR,
(errcode(EINVAL),
errmsg("cannot repack specific table(s) in all databases")));
if (schema_list.head)
ereport(ERROR,
(errcode(EINVAL),
errmsg("cannot repack specific schema(s) in all databases")));
repack_all_databases(orderby);
}
else
@ -563,12 +574,14 @@ repack_one_database(const char *orderby, char *errbuf, size_t errsize)
const char **params = NULL;
int iparam = 0;
size_t num_tables;
size_t num_schemas;
size_t num_params;
num_tables = simple_string_list_size(table_list);
num_schemas = simple_string_list_size(schema_list);
/* 1st param is the user-specified tablespace */
num_params = num_tables + 1;
num_params = num_tables + num_schemas + 1;
params = pgut_malloc(num_params * sizeof(char *));
initStringInfo(&sql);
@ -604,6 +617,19 @@ repack_one_database(const char *orderby, char *errbuf, size_t errsize)
}
appendStringInfoString(&sql, ")");
}
else if (num_schemas)
{
appendStringInfoString(&sql, "schemaname IN (");
for (cell = schema_list.head; cell; cell = cell->next)
{
/* Construct schema name placeholders to be used by PQexecParams */
appendStringInfo(&sql, "$%d", iparam + 1);
params[iparam++] = cell->val;
if (cell->next)
appendStringInfoString(&sql, ", ");
}
appendStringInfoString(&sql, ")");
}
else
{
appendStringInfoString(&sql, "pkid IS NOT NULL");
@ -645,6 +671,7 @@ repack_one_database(const char *orderby, char *errbuf, size_t errsize)
table.target_oid = getoid(res, i, c++);
table.target_toast = getoid(res, i, c++);
table.target_tidx = getoid(res, i, c++);
c++; // Skip schemaname
table.pkid = getoid(res, i, c++);
table.ckid = getoid(res, i, c++);
@ -1671,6 +1698,7 @@ repack_table_indexes(PGresult *index_details)
for (i = 0; i < num; i++)
{
char *isvalid = getstr(index_details, i, 2);
char *idx_name = getstr(index_details, i, 0);
if (isvalid[0] == 't')
{
@ -1683,7 +1711,7 @@ repack_table_indexes(PGresult *index_details)
"WHERE pgc.relname = 'index_%u' "
"AND nsp.nspname = $1", index);
params[0] = schema_name;
elog(INFO, "repacking index \"%s\".\"index_%u\"", schema_name, index);
elog(INFO, "repacking index \"%s\".\"%s\"", schema_name, idx_name);
res = execute(sql.data, 1, params);
if (PQresultStatus(res) != PGRES_TUPLES_OK)
{
@ -1837,6 +1865,10 @@ repack_all_indexes(char *errbuf, size_t errsize)
if (!preliminary_checks(errbuf, errsize))
goto cleanup;
/* XXX: tighten these ORDER BYs to avoid intermittent installcheck
* failures due to differently-ordered results for some of the
* --only-indexes tests.
*/
if (r_index.head)
{
appendStringInfoString(&sql,
@ -1883,8 +1915,6 @@ repack_all_indexes(char *errbuf, size_t errsize)
if(table_list.head)
elog(INFO, "repacking indexes of \"%s\"", cell->val);
else
elog(INFO, "repacking \"%s\"", cell->val);
if (!repack_table_indexes(res))
elog(WARNING, "repack failed for \"%s\"", cell->val);
@ -1913,6 +1943,7 @@ pgut_help(bool details)
printf("Options:\n");
printf(" -a, --all repack all databases\n");
printf(" -t, --table=TABLE repack specific table only\n");
printf(" -c, --schema=SCHEMA repack tables in specific schema only\n");
printf(" -s, --tablespace=TBLSPC move repacked tables to a new tablespace\n");
printf(" -S, --moveidx move repacked indexes to TBLSPC too\n");
printf(" -o, --order-by=COLUMNS order by columns instead of cluster keys\n");

View File

@ -117,6 +117,7 @@ The following options can be specified in ``OPTIONS``.
Options:
-a, --all repack all databases
-t, --table=TABLE repack specific table only
-c, --schema=SCHEMA repack tables in specific schema only
-s, --tablespace=TBLSPC move repacked tables to a new tablespace
-S, --moveidx move repacked indexes to *TBLSPC* too
-o, --order-by=COLUMNS order by columns instead of cluster keys
@ -155,6 +156,11 @@ Reorg Options
reorganized by writing multiple ``-t`` switches. By default, all eligible
tables in the target databases are reorganized.
``-c``, ``--schema``
Repack the tables in the specified schema(s) only. Multiple schemas may
be repacked by writing multiple ``-c`` switches. May be used in
conjunction with ``--tablespace`` to move tables to a different tablespace.
``-o COLUMNS [,...]``, ``--order-by=COLUMNS [,...]``
Perform an online CLUSTER ordered by the specified columns.
@ -438,6 +444,7 @@ Releases
* pg_repack 1.3
* Added ``--dry-run`` to do a dry run.
* Added ``--schema`` to repack only the specified schema.
* pg_repack 1.2

View File

@ -173,6 +173,7 @@ CREATE VIEW repack.tables AS
R.oid AS relid,
R.reltoastrelid AS reltoastrelid,
CASE WHEN R.reltoastrelid = 0 THEN 0 ELSE (SELECT reltoastidxid FROM pg_class WHERE oid = R.reltoastrelid) END AS reltoastidxid,
N.nspname AS schemaname,
PK.indexrelid AS pkid,
CK.indexrelid AS ckid,
repack.get_create_index_type(PK.indexrelid, 'repack.pk_' || R.oid) AS create_pktype,

View File

@ -325,6 +325,8 @@ INFO: repacking table "tbl_pk_uk"
-- => OK
\! pg_repack --dbname=contrib_regression --table=tbl_pk_uk --only-indexes
INFO: repacking indexes of "tbl_pk_uk"
INFO: repacking index "public"."tbl_pk_uk_pkey"
INFO: repacking index "public"."tbl_pk_uk_col2_col1_key"
-- => OK
\! pg_repack --dbname=contrib_regression --table=tbl_nn_puk
WARNING: relation "tbl_nn_puk" must have a primary key or not-null unique keys
@ -361,3 +363,27 @@ INFO: repacking table "trg4"
\! pg_repack --dbname=contrib_regression --table=tbl_cluster --dry-run
INFO: Dry run enabled, not executing repack
INFO: repacking table "tbl_cluster"
-- Test --schema
--
CREATE SCHEMA test_schema1;
CREATE TABLE test_schema1.tbl1 (id INTEGER PRIMARY KEY);
CREATE TABLE test_schema1.tbl2 (id INTEGER PRIMARY KEY);
CREATE SCHEMA test_schema2;
CREATE TABLE test_schema2.tbl1 (id INTEGER PRIMARY KEY);
CREATE TABLE test_schema2.tbl2 (id INTEGER PRIMARY KEY);
-- => OK
\! pg_repack --dbname=contrib_regression --schema=test_schema1
INFO: repacking table "test_schema1.tbl1"
INFO: repacking table "test_schema1.tbl2"
-- => OK
\! pg_repack --dbname=contrib_regression --schema=test_schema1 --schema=test_schema2
INFO: repacking table "test_schema2.tbl1"
INFO: repacking table "test_schema2.tbl2"
INFO: repacking table "test_schema1.tbl2"
INFO: repacking table "test_schema1.tbl1"
-- => ERROR
\! pg_repack --dbname=contrib_regression --schema=test_schema1 --table=tbl1
ERROR: cannot repack specific table(s) in schema, use schema.table notation instead
-- => ERROR
\! pg_repack --dbname=contrib_regression --all --schema=test_schema1
ERROR: cannot repack specific schema(s) in all databases

View File

@ -135,6 +135,9 @@ INFO: repacking table "testts1"
--move all indexes of the table to a tablespace
\! pg_repack --dbname=contrib_regression --table=testts1 --only-indexes --tablespace=testts
INFO: repacking indexes of "testts1"
INFO: repacking index "public"."testts1_pkey"
INFO: repacking index "public"."testts1_partial_idx"
INFO: repacking index "public"."testts1_with_idx"
SELECT relname, spcname
FROM pg_class JOIN pg_tablespace ts ON ts.oid = reltablespace
WHERE relname ~ '^testts1'
@ -149,6 +152,9 @@ ORDER BY relname;
--all indexes of tablespace remain in same tablespace
\! pg_repack --dbname=contrib_regression --table=testts1 --only-indexes
INFO: repacking indexes of "testts1"
INFO: repacking index "public"."testts1_pkey"
INFO: repacking index "public"."testts1_partial_idx"
INFO: repacking index "public"."testts1_with_idx"
SELECT relname, spcname
FROM pg_class JOIN pg_tablespace ts ON ts.oid = reltablespace
WHERE relname ~ '^testts1'
@ -163,6 +169,9 @@ ORDER BY relname;
--move all indexes of the table to pg_default
\! pg_repack --dbname=contrib_regression --table=testts1 --only-indexes --tablespace=pg_default
INFO: repacking indexes of "testts1"
INFO: repacking index "public"."testts1_pkey"
INFO: repacking index "public"."testts1_partial_idx"
INFO: repacking index "public"."testts1_with_idx"
SELECT relname, spcname
FROM pg_class JOIN pg_tablespace ts ON ts.oid = reltablespace
WHERE relname ~ '^testts1'
@ -173,7 +182,7 @@ ORDER BY relname;
--move one index to a tablespace
\! pg_repack --dbname=contrib_regression --index=testts1_pkey --tablespace=testts
INFO: repacking "testts1_pkey"
INFO: repacking index "public"."testts1_pkey"
SELECT relname, spcname
FROM pg_class JOIN pg_tablespace ts ON ts.oid = reltablespace
WHERE relname ~ '^testts1'
@ -185,7 +194,7 @@ ORDER BY relname;
--index tablespace stays as is
\! pg_repack --dbname=contrib_regression --index=testts1_pkey
INFO: repacking "testts1_pkey"
INFO: repacking index "public"."testts1_pkey"
SELECT relname, spcname
FROM pg_class JOIN pg_tablespace ts ON ts.oid = reltablespace
WHERE relname ~ '^testts1'
@ -197,7 +206,7 @@ ORDER BY relname;
--move index to pg_default
\! pg_repack --dbname=contrib_regression --index=testts1_pkey --tablespace=pg_default
INFO: repacking "testts1_pkey"
INFO: repacking index "public"."testts1_pkey"
SELECT relname, spcname
FROM pg_class JOIN pg_tablespace ts ON ts.oid = reltablespace
WHERE relname ~ '^testts1'
@ -208,8 +217,8 @@ ORDER BY relname;
--using multiple --index option
\! pg_repack --dbname=contrib_regression --index=testts1_pkey --index=testts1_with_idx --tablespace=testts
INFO: repacking "testts1_pkey"
INFO: repacking "testts1_with_idx"
INFO: repacking index "public"."testts1_pkey"
INFO: repacking index "public"."testts1_with_idx"
SELECT relname, spcname
FROM pg_class JOIN pg_tablespace ts ON ts.oid = reltablespace
WHERE relname ~ '^testts1'

View File

@ -213,3 +213,20 @@ CREATE TRIGGER zzzzzz AFTER UPDATE ON trg4 FOR EACH ROW EXECUTE PROCEDURE trgtes
-- Dry run
--
\! pg_repack --dbname=contrib_regression --table=tbl_cluster --dry-run
-- Test --schema
--
CREATE SCHEMA test_schema1;
CREATE TABLE test_schema1.tbl1 (id INTEGER PRIMARY KEY);
CREATE TABLE test_schema1.tbl2 (id INTEGER PRIMARY KEY);
CREATE SCHEMA test_schema2;
CREATE TABLE test_schema2.tbl1 (id INTEGER PRIMARY KEY);
CREATE TABLE test_schema2.tbl2 (id INTEGER PRIMARY KEY);
-- => OK
\! pg_repack --dbname=contrib_regression --schema=test_schema1
-- => OK
\! pg_repack --dbname=contrib_regression --schema=test_schema1 --schema=test_schema2
-- => ERROR
\! pg_repack --dbname=contrib_regression --schema=test_schema1 --table=tbl1
-- => ERROR
\! pg_repack --dbname=contrib_regression --all --schema=test_schema1