Re: [HACKERS] Restrict concurrent update/delete with UPDATE of partition key

On Fri, Jan 12, 2018 at 11:43 AM, amul sul <sulamul(at)gmail(dot)com> wrote:
>
> Thanks for looking at this thread, attached herewith an updated patch rebase on
> 'UPDATE of partition key v35' patch[1].
>

ExecDelete(mtstate, tupleid, oldtuple, planSlot, epqstate, estate,
- &tuple_deleted, false, false);
+ &tuple_deleted, false, false, true);

/*
* For some reason if DELETE didn't happen (e.g. trigger prevented
@@ -1292,6 +1299,11 @@ lreplace:;
ereport(ERROR,
(errcode(ERRCODE_T_R_SERIALIZATION_FAILURE),
errmsg("could not serialize access due to concurrent update")));
+ if (!BlockNumberIsValid(BlockIdGetBlockNumber(&((hufd.ctid).ip_blkid))))
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("tuple to be locked was already moved to another partition
due to concurrent update")));

I have asked to change the message "tuple to be updated .." after
heap_lock_tuple call not in nodeModifyTable.c, so please revert the
message in nodeModifyTable.c.

Have you verified the changes in execReplication.c in some way? It is
not clear to me how do you ensure to set the special value
(InvalidBlockNumber) in CTID for delete operation via logical
replication? The logical replication worker uses function
ExecSimpleRelationDelete to perform Delete and there is no way it can
pass the correct value of row_moved in heap_delete. Am I missing
something due to which we don't need to do this?

--
With Regards,
Amit Kapila.
EnterpriseDB: http://www.enterprisedb.com