database/postgres
1
0
Fork 0
mirror of https://github.com/postgres/postgres.git synced 2025-09-02 04:21:28 +03:00
Code Activity

Implement genuine serializable isolation level.

Browse Source 
Until now, our Serializable mode has in fact been what's called Snapshot
Isolation, which allows some anomalies that could not occur in any
serialized ordering of the transactions. This patch fixes that using a
method called Serializable Snapshot Isolation, based on research papers by
Michael J. Cahill (see README-SSI for full references). In Serializable
Snapshot Isolation, transactions run like they do in Snapshot Isolation,
but a predicate lock manager observes the reads and writes performed and
aborts transactions if it detects that an anomaly might occur. This method
produces some false positives, ie. it sometimes aborts transactions even
though there is no anomaly.

To track reads we implement predicate locking, see storage/lmgr/predicate.c.
Whenever a tuple is read, a predicate lock is acquired on the tuple. Shared
memory is finite, so when a transaction takes many tuple-level locks on a
page, the locks are promoted to a single page-level lock, and further to a
single relation level lock if necessary. To lock key values with no matching
tuple, a sequential scan always takes a relation-level lock, and an index
scan acquires a page-level lock that covers the search key, whether or not
there are any matching keys at the moment.

A predicate lock doesn't conflict with any regular locks or with another
predicate locks in the normal sense. They're only used by the predicate lock
manager to detect the danger of anomalies. Only serializable transactions
participate in predicate locking, so there should be no extra overhead for
for other transactions.

Predicate locks can't be released at commit, but must be remembered until
all the transactions that overlapped with it have completed. That means that
we need to remember an unbounded amount of predicate locks, so we apply a
lossy but conservative method of tracking locks for committed transactions.
If we run short of shared memory, we overflow to a new "pg_serial" SLRU
pool.

We don't currently allow Serializable transactions in Hot Standby mode.
That would be hard, because even read-only transactions can cause anomalies
that wouldn't otherwise occur.

Serializable isolation mode now means the new fully serializable level.
Repeatable Read gives you the old Snapshot Isolation level that we have
always had.

Kevin Grittner and Dan Ports, reviewed by Jeff Davis, Heikki Linnakangas and
Anssi Kääriäinen
This commit is contained in:
Heikki Linnakangas
2011-02-07 23:46:51 +02:00
parent c18f51da17
commit dafaa3efb7
90 changed files with 14995 additions and 271 deletions
Download Patch File Download Diff File Expand all files Collapse all files

35
src/bin/pg_dump/pg_dump.c
View File

@@ -11,14 +11,14 @@
* script that reproduces the schema in terms of SQL that is understood
* by PostgreSQL
*
* Note that pg_dump runs in a serializable transaction, so it sees a
* consistent snapshot of the database including system catalogs.
* However, it relies in part on various specialized backend functions
* like pg_get_indexdef(), and those things tend to run on SnapshotNow
* time, ie they look at the currently committed state. So it is
* possible to get 'cache lookup failed' error if someone performs DDL
* changes while a dump is happening. The window for this sort of thing
* is from the beginning of the serializable transaction to
* Note that pg_dump runs in a transaction-snapshot mode transaction,
* so it sees a consistent snapshot of the database including system
* catalogs. However, it relies in part on various specialized backend
* functions like pg_get_indexdef(), and those things tend to run on
* SnapshotNow time, ie they look at the currently committed state. So
* it is possible to get 'cache lookup failed' error if someone
* performs DDL changes while a dump is happening. The window for this
* sort of thing is from the acquisition of the transaction snapshot to
* getSchemaData() (when pg_dump acquires AccessShareLock on every
* table it intends to dump). It isn't very large, but it can happen.
*
@@ -135,6 +135,7 @@ static int dump_inserts = 0;
static int column_inserts = 0;
static int no_security_label = 0;
static int no_unlogged_table_data = 0;
static int serializable_deferrable = 0;
static void help(const char *progname);
@@ -318,6 +319,7 @@ main(int argc, char **argv)
{"no-tablespaces", no_argument, &outputNoTablespaces, 1},
{"quote-all-identifiers", no_argument, &quote_all_identifiers, 1},
{"role", required_argument, NULL, 3},
{"serializable-deferrable", no_argument, &serializable_deferrable, 1},
{"use-set-session-authorization", no_argument, &use_setsessauth, 1},
{"no-security-label", no_argument, &no_security_label, 1},
{"no-unlogged-table-data", no_argument, &no_unlogged_table_data, 1},
@@ -669,11 +671,21 @@ main(int argc, char **argv)
no_security_label = 1;
/*
* Start serializable transaction to dump consistent data.
* Start transaction-snapshot mode transaction to dump consistent data.
*/
do_sql_command(g_conn, "BEGIN");
do_sql_command(g_conn, "SET TRANSACTION ISOLATION LEVEL SERIALIZABLE");
if (g_fout->remoteVersion >= 90100)
{
if (serializable_deferrable)
do_sql_command(g_conn,
"SET TRANSACTION ISOLATION LEVEL SERIALIZABLE, "
"READ ONLY, DEFERRABLE");
else
do_sql_command(g_conn,
"SET TRANSACTION ISOLATION LEVEL REPEATABLE READ");
}
else
do_sql_command(g_conn, "SET TRANSACTION ISOLATION LEVEL SERIALIZABLE");
/* Select the appropriate subquery to convert user IDs to names */
if (g_fout->remoteVersion >= 80100)
@@ -864,6 +876,7 @@ help(const char *progname)
printf(_(" --disable-triggers disable triggers during data-only restore\n"));
printf(_(" --no-tablespaces do not dump tablespace assignments\n"));
printf(_(" --quote-all-identifiers quote all identifiers, even if not keywords\n"));
printf(_(" --serializable-deferrable wait until the dump can run without anomalies\n"));
printf(_(" --role=ROLENAME do SET ROLE before dump\n"));
printf(_(" --no-security-label do not dump security label assignments\n"));
printf(_(" --no-unlogged-table-data do not dump unlogged table data\n"));