| Bug #15718 | mysql_real_connect() using named pipe doesn't honor wait_timeout | ||
|---|---|---|---|
| Submitted: | 13 Dec 2005 21:22 | Modified: | 20 Jan 2006 22:40 |
| Reporter: | Edmund Mierzwa | Email Updates: | |
| Status: | Closed | Impact on me: | |
| Category: | MySQL Server: Documentation | Severity: | S3 (Non-critical) |
| Version: | 5.0.16-nt-max | OS: | Windows (XP, with SP2) |
| Assigned to: | Paul DuBois | CPU Architecture: | Any |
[13 Dec 2005 23:24]
MySQL Verification Team
Changing to documentation category, on Windows the timeout are based on TCP/IP socket's timeouts that is the reason why behaves in this way. Thank you for the bug report.
[20 Jan 2006 22:07]
MySQL Verification Team
Paul,
The bug report and my comments are limited just for our Windows servers.
The timeouts were implemented using TCP/IP Windows sockets how you
can see below in the piece of code extracted from mysqld.cc 4.1:
#ifdef __WIN__
/* Set default wait_timeout */
ulong wait_timeout= global_system_variables.net_wait_timeout * 1000;
(void) setsockopt(new_sock, SOL_SOCKET, SO_RCVTIMEO, (char*)&wait_timeout,
sizeof(wait_timeout));
#endif
so named pipes and shared memory not work with these timeouts just
connections using TCP/IP.
There is a bug report, sorry don't remember the number submited with
a patch how to implement the same kind of timeout on Unices.
[20 Jan 2006 22:40]
Paul DuBois
Thank you for your bug report. This issue has been addressed in the documentation. The updated documentation will appear on our website shortly, and will be included in the next release of the relevant product(s). Additional info: I'll update the description of wait_timeout here to indicate that it applies only to TCP/IP connections: http://dev.mysql.com/doc/refman/5.0/en/server-system-variables.html
Description: Restarted server with wait_timeout=4. Then forcing named pipe connection mysql_real_connect( ,".", ) and my idle application does not timeout after exceeding timeout value. If I change "." to "localhost" the app will timeout at the set duration. Client and server are both at 5.0.16-nt-max. I could not find any documentation indicating this behaviour or active bugs or mentioned in the windows list. How to repeat: [mysqld] skip-bdb skip-innodb skip-ndbcluster enable-named-pipe basedir =c:/mysql_5.0 datadir =c:/mysql_5.0/data language=c:/mysql_5.0/share/english sql-mode=ANSI_QUOTES,ERROR_FOR_DIVISION_BY_ZERO,IGNORE_SPACE,NO_AUTO_CREATE_USER,NO_BACKSLASH_ESCAPES,NO_ZERO_DATE,NO_ZERO_IN_DATE,PIPES_AS_CONCAT,REAL_AS_FLOAT,STRICT_ALL_TABLES set-variable = wait_timeout=4 set-variable = interactive_timeout=5 MYSQL * db_connect(const char *dbname) { my_bool my_true = true; MYSQL *db = mysql_init(NULL); if (!db) die(db, "mysql_init failed: no memory"); if (!mysql_real_connect(db, ".", "oven", "xxx", dbname, 0, NULL, CLIENT_MULTI_STATEMENTS )) die(db, "mysql_real_connect() failed: %s", mysql_error(db)); return db; } Suggested fix: Behavior for TCPIP and named pipe connection timeouts should be the same.