Wed Dec 14 16:15:26 PST 2005
- Previous message: [Slony1-commit] By wieck: Rebuild listen entries also on change of data provider for and
- Next message: [Slony1-commit] By wieck: Remove the use of the mutex lock over PQconnect() if we are
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Log Message:
-----------
With the new watchdog behaviour, the macro's slon_retry() and
slon_restart() should not return at all any more. If a thread
has decided that a complete slon restart is in order, it now
bails out. The watchdog will clean up the mess.
Jan
Modified Files:
--------------
slony1-engine/src/slon:
slon.h (r1.57 -> r1.58)
-------------- next part --------------
Index: slon.h
===================================================================
RCS file: /usr/local/cvsroot/slony1/slony1-engine/src/slon/slon.h,v
retrieving revision 1.57
retrieving revision 1.58
diff -Lsrc/slon/slon.h -Lsrc/slon/slon.h -u -w -r1.57 -r1.58
--- src/slon/slon.h
+++ src/slon/slon.h
@@ -369,6 +369,7 @@
slon_watchdog_pid = -1; \
} \
pthread_mutex_unlock(&slon_watchdog_lock); \
+ pthread_exit(NULL); \
} while (0)
#define slon_retry() \
do { \
@@ -379,6 +380,7 @@
slon_watchdog_pid = -1; \
} \
pthread_mutex_unlock(&slon_watchdog_lock); \
+ pthread_exit(NULL); \
} while (0)
#else /* WIN32 */
/* On win32, we currently just bail out and let the service control manager
- Previous message: [Slony1-commit] By wieck: Rebuild listen entries also on change of data provider for and
- Next message: [Slony1-commit] By wieck: Remove the use of the mutex lock over PQconnect() if we are
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Slony1-commit mailing list