Hi,
On larger, multi-socket, machines, startup takes a fair bit of time. As
I was profiling anyway I looked into it and noticed that just about all
of it is spent in LWLockAssign() called by InitBufferPool(). Starting
with shared_buffers=48GB on the server Nate Boley provided, takes about
12 seconds. Nearly all of it spent taking the ShmemLock spinlock.
Simply modifying LWLockAssign() to not take the spinlock when
!IsUnderPostmaster speeds it up to 2 seconds. While certainly not making
LWLockAssign() prettier it seems enough of a speedup to be worthwile
nonetheless.
Since this code is also hit when do an emergency restart, I'd say it has
practical relevance...
Greetings,
Andres Freund
--
Andres Freund http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services
diff --git a/src/backend/storage/lmgr/lwlock.c b/src/backend/storage/lmgr/lwlock.c
index 0679eb5..10ee364 100644
--- a/src/backend/storage/lmgr/lwlock.c
+++ b/src/backend/storage/lmgr/lwlock.c
@@ -451,9 +451,11 @@ CreateLWLocks(void)
* LWLockAssign - assign a dynamically-allocated LWLock number
*
* We interlock this using the same spinlock that is used to protect
- * ShmemAlloc(). Interlocking is not really necessary during postmaster
- * startup, but it is needed if any user-defined code tries to allocate
- * LWLocks after startup.
+ * ShmemAlloc(). Interlocking is not necessary during postmaster startup, but
+ * it is needed if any user-defined code tries to allocate LWLocks after
+ * startup. Since we allocate large amounts of LWLocks for the buffer pool, we
+ * avoid taking the spinlock when not needed, as it has shown to slowdown
+ * startup considerably.
*/
LWLock *
LWLockAssign(void)
@@ -464,14 +466,17 @@ LWLockAssign(void)
volatile int *LWLockCounter;
LWLockCounter = (int *) ((char *) MainLWLockArray - 3 * sizeof(int));
- SpinLockAcquire(ShmemLock);
+ if (IsUnderPostmaster)
+ SpinLockAcquire(ShmemLock);
if (LWLockCounter[0] >= LWLockCounter[1])
{
- SpinLockRelease(ShmemLock);
+ if (IsUnderPostmaster)
+ SpinLockRelease(ShmemLock);
elog(ERROR, "no more LWLocks available");
}
result = &MainLWLockArray[LWLockCounter[0]++].lock;
- SpinLockRelease(ShmemLock);
+ if (IsUnderPostmaster)
+ SpinLockRelease(ShmemLock);
return result;
}
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
