Обсуждение: behaviour change - default_tablesapce + partition table
Hi,
Consider the below test:
create tablespace mytbs location '/home/rushabh/mywork/workspace/pg/';
create table test ( a int , b int ) partition by list (a);
set default_tablespace to mytbs;
create table test_p1 partition of test for values in (1);
create tablespace mytbs location '/home/rushabh/mywork/workspace/pg/';
create table test ( a int , b int ) partition by list (a);
set default_tablespace to mytbs;
create table test_p1 partition of test for values in (1);
In the above test, after the setting the default_tablespace I am creating
a partition table and expecting that to get created under "mytbs" tablespace.
But that is not the case:
postgres@66247=#select relname, reltablespace from pg_class where relname = 'test_p1';
relname | reltablespace
---------+---------------
test_p1 | 0
(1 row)
with below commit.
Author: Alvaro Herrera <alvherre@alvh.no-ip.org>
Date: Thu Apr 25 10:20:23 2019 -0400
Fix tablespace inheritance for partitioned rels
Commit ca4103025dfe left a few loose ends. The most important one
(broken pg_dump output) is already fixed by virtue of commit
3b23552ad8bb, but some things remained:
I don't think that new behaviour is intended and if it's an intended change
than need to fix pg_dump as well - other wise lets say,
1) create the above test on v11
2) take a dump
3) restore on v12
will end up partition into wrong tablesapce.
Looking at the commit changes, it seems like at condition when no other
tablespace is specified, we default the tablespace to the parent partitioned
table's.
else if (stmt->partbound)
{
/*
* For partitions, when no other tablespace is specified, we default
* the tablespace to the parent partitioned table's.
*/
Assert(list_length(inheritOids) == 1);
tablespaceId = get_rel_tablespace(linitial_oid(inheritOids));
}
tablespace is specified, we default the tablespace to the parent partitioned
table's.
else if (stmt->partbound)
{
/*
* For partitions, when no other tablespace is specified, we default
* the tablespace to the parent partitioned table's.
*/
Assert(list_length(inheritOids) == 1);
tablespaceId = get_rel_tablespace(linitial_oid(inheritOids));
}
But here it doesn't consider the default_tablespace if the parent partitioned
tablespace is an InvalidOid (which was the care before this commit).
PFA patch to fix the same.
Thanks,
Вложения
Agree that this behavior change seems unintentional. On 2019/05/17 12:40, Rushabh Lathia wrote: > Looking at the commit changes, it seems like at condition when no other > tablespace is specified, we default the tablespace to the parent partitioned > table's. > > else if (stmt->partbound) > { > /* > * For partitions, when no other tablespace is specified, we default > * the tablespace to the parent partitioned table's. > */ > Assert(list_length(inheritOids) == 1); > tablespaceId = get_rel_tablespace(linitial_oid(inheritOids)); > } > > But here it doesn't consider the default_tablespace if the parent > partitioned > tablespace is an InvalidOid (which was the care before this commit). > > PFA patch to fix the same. + + if (!OidIsValid(tablespaceId)) + tablespaceId = GetDefaultTablespace(stmt->relation->relpersistence, partitioned); } else tablespaceId = GetDefaultTablespace(stmt->relation->relpersistence, Why not change it like this instead: @@ -681,7 +681,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId, Assert(list_length(inheritOids) == 1); tablespaceId = get_rel_tablespace(linitial_oid(inheritOids)); } - else + + if (!OidIsValid(tablespaceId)) tablespaceId = GetDefaultTablespace(stmt->relation->relpersistence, partitioned); Thanks, Amit
On 2019/05/17 12:40, Rushabh Lathia wrote: > Hi, > > Consider the below test: > > create tablespace mytbs location '/home/rushabh/mywork/workspace/pg/'; > create table test ( a int , b int ) partition by list (a); > > set default_tablespace to mytbs; > create table test_p1 partition of test for values in (1); > > In the above test, after the setting the default_tablespace I am creating > a partition table and expecting that to get created under "mytbs" > tablespace. > > But that is not the case: > > postgres@66247=#select relname, reltablespace from pg_class where relname = > 'test_p1'; > relname | reltablespace > ---------+--------------- > test_p1 | 0 > (1 row) > > I noticed the behaviour change for default_tablespace with partition table > with below commit. > > commit 87259588d0ab0b8e742e30596afa7ae25caadb18 > Author: Alvaro Herrera <alvherre@alvh.no-ip.org> > Date: Thu Apr 25 10:20:23 2019 -0400 > > Fix tablespace inheritance for partitioned rels > > Commit ca4103025dfe left a few loose ends. The most important one > (broken pg_dump output) is already fixed by virtue of commit > 3b23552ad8bb, but some things remained: > > I don't think that new behaviour is intended Should we add this to open items? Thanks, Amit
On Fri, May 17, 2019 at 10:30 AM Amit Langote <Langote_Amit_f8@lab.ntt.co.jp> wrote:
Agree that this behavior change seems unintentional.
On 2019/05/17 12:40, Rushabh Lathia wrote:
> Looking at the commit changes, it seems like at condition when no other
> tablespace is specified, we default the tablespace to the parent partitioned
> table's.
>
> else if (stmt->partbound)
> {
> /*
> * For partitions, when no other tablespace is specified, we default
> * the tablespace to the parent partitioned table's.
> */
> Assert(list_length(inheritOids) == 1);
> tablespaceId = get_rel_tablespace(linitial_oid(inheritOids));
> }
>
> But here it doesn't consider the default_tablespace if the parent
> partitioned
> tablespace is an InvalidOid (which was the care before this commit).
>
> PFA patch to fix the same.
+
+ if (!OidIsValid(tablespaceId))
+ tablespaceId = GetDefaultTablespace(stmt->relation->relpersistence,
partitioned);
}
else
tablespaceId = GetDefaultTablespace(stmt->relation->relpersistence,
Why not change it like this instead:
@@ -681,7 +681,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid
ownerId,
Assert(list_length(inheritOids) == 1);
tablespaceId = get_rel_tablespace(linitial_oid(inheritOids));
}
- else
+
+ if (!OidIsValid(tablespaceId))
tablespaceId = GetDefaultTablespace(stmt->relation->relpersistence,
partitioned);
Yes, sure we can do that. Here is the patch for the same.
Rushabh Lathia
Вложения
On 2019/05/20 13:42, Rushabh Lathia wrote: > On Fri, May 17, 2019 at 10:30 AM Amit Langote <Langote_Amit_f8@lab.ntt.co.jp> >> Why not change it like this instead: >> >> @@ -681,7 +681,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid >> ownerId, >> Assert(list_length(inheritOids) == 1); >> tablespaceId = get_rel_tablespace(linitial_oid(inheritOids)); >> } >> - else >> + >> + if (!OidIsValid(tablespaceId)) >> tablespaceId = >> GetDefaultTablespace(stmt->relation->relpersistence, >> partitioned); > > Yes, sure we can do that. Here is the patch for the same. Thanks Rushabh. Regards, Amit
On 2019-May-20, Amit Langote wrote: > Should we add this to open items? Yeah. I'm AFK this week, but can handle it afterwards. The fix already missed beta1, so I don't think there's a problem with taking a little bit longer. -- Álvaro Herrera https://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
On 2019/05/21 11:29, Alvaro Herrera wrote: > On 2019-May-20, Amit Langote wrote: > >> Should we add this to open items? > > Yeah. I'm AFK this week, but can handle it afterwards. The fix already > missed beta1, so I don't think there's a problem with taking a little > bit longer. OK, added. Thanks, Amit
On 2019-May-20, Alvaro Herrera wrote: > On 2019-May-20, Amit Langote wrote: > > > Should we add this to open items? > > Yeah. I'm AFK this week, but can handle it afterwards. The fix already > missed beta1, so I don't think there's a problem with taking a little > bit longer. Here's my proposed patch. I changed/expanded the tests a little bit to ensure more complete coverage. -- Álvaro Herrera https://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
Вложения
On 2019-Jun-06, Alvaro Herrera wrote: > Here's my proposed patch. I changed/expanded the tests a little bit to > ensure more complete coverage. Well, revise the comments a little bit. -- Álvaro Herrera https://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
Вложения
Hi Alvaro, On Fri, Jun 7, 2019 at 8:02 AM Alvaro Herrera <alvherre@2ndquadrant.com> wrote: > On 2019-Jun-06, Alvaro Herrera wrote: > > > Here's my proposed patch. I changed/expanded the tests a little bit to > > ensure more complete coverage. > > Well, revise the comments a little bit. Thanks for adding the tests. Looks good. Regards, Amit
Hi Amit, Rushabh, On 2019-Jun-07, Amit Langote wrote: > Thanks for adding the tests. Looks good. Thanks Amit, pushed now. Marking open item as done. -- Álvaro Herrera https://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services