Fix typo s/exlusive/exclusive/

pull/5006/head
Hanefi Onaldi 2021-12-22 21:24:58 +03:00 committed by Hanefi Onaldi
parent 1af8ca8f7c
commit 76176caea7
3 changed files with 4 additions and 4 deletions

View File

@ -2660,7 +2660,7 @@ upgrade_columnar_storage(PG_FUNCTION_ARGS)
* ACCESS EXCLUSIVE LOCK is not required by the low-level routines, so we
* can take only an ACCESS SHARE LOCK. But all access to non-current
* columnar tables will fail anyway, so it's better to take ACCESS
* EXLUSIVE LOCK now.
* EXCLUSIVE LOCK now.
*/
Relation rel = table_open(relid, AccessExclusiveLock);
if (!IsColumnarTableAmTable(relid))
@ -2696,7 +2696,7 @@ downgrade_columnar_storage(PG_FUNCTION_ARGS)
* ACCESS EXCLUSIVE LOCK is not required by the low-level routines, so we
* can take only an ACCESS SHARE LOCK. But all access to non-current
* columnar tables will fail anyway, so it's better to take ACCESS
* EXLUSIVE LOCK now.
* EXCLUSIVE LOCK now.
*/
Relation rel = table_open(relid, AccessExclusiveLock);
if (!IsColumnarTableAmTable(relid))

View File

@ -117,7 +117,7 @@ AcquireExecutorShardLocksForRelationRowLockList(List *relationRowLockList)
* We have selected these lock types according to conflict table given in the
* Postgres documentation. It is given that FOR UPDATE and FOR NO KEY UPDATE
* must be conflict with each other modify command. By getting ExlcusiveLock
* we guarantee that. Note that, getting ExlusiveLock does not mimic the
* we guarantee that. Note that, getting ExclusiveLock does not mimic the
* behaviour of Postgres exactly. Getting row lock with FOR NO KEY UPDATE and
* FOR KEY SHARE do not conflict in Postgres, yet they block each other in
* our implementation. Since FOR SHARE and FOR KEY SHARE does not conflict

View File

@ -183,7 +183,7 @@ RecoverWorkerTransactions(WorkerNode *workerNode)
* distributed transactions.
*
* We could avoid this by temporarily blocking new prepared transactions
* from being created by taking an ExlusiveLock on pg_dist_transaction.
* from being created by taking an ExclusiveLock on pg_dist_transaction.
* However, this hurts write performance, so instead we avoid blocking
* by consulting the list of active distributed transactions, and follow
* a carefully chosen order to avoid race conditions: