mysterious "can 't connect " error message in our logs(2)

Mailing List

Home

MySQL General - General MySQL discussion

MaxDB - Everything about MaxDB, formerly known as SAP DB

MySQL on Win32 - Runing MySQL on Windows 9x/Me/NT/2000/XP

Java Help - Mostly related to the MySQL Connector/J driver

ODBC - ODBC with the MySQL Connector/ODBC driver

Perl - Perl support for MySQL with DBI and DBD::mysql

MySQL++ - Programming with the C++ API to MySQL

Subjects

•	mysql openssl Question
•	mysql:it 's a db not a dbms how it 's possible?!
•	Does the binary log enabling affect the MySQL performances?
•	Strange behavior, Table Level Permission
•	FULLTEXT query format question
•	Preventing Duplicate Entries
•	Comparing and writing out BLOBS
•	Executing MySQL Commands From Within C Program
•	Can 't access mysql after kernel upgrade
•	Mysql 4 0 Oracle Stored Procedure Trigger Conversion
•	Downgrade Mysql from 4 to 3 23
•	MySQL Cluster Software
•	mysql test 4 1 fails with the gis test
•	ERROR 2002: Can 't connect to local MySQL server through socket
•	Getting Identity after INSERT
•	Update one field with more fields from another table
•	ERROR 1045: Access denied for user: 'root@localhost ' (Using password: NO)
•	mysql have same function mthod as Oracle decode()

mysterious "can 't connect " error message in our logs(2)

mysterious "can 't connect " error message in our logs(2)

2004-03-22 - By Jigal van Hemert

Back

Reply: 1 2 3

The continuing saga:

> Can 't connect to MySQL server on '192.168.13.205 ' (4)
Our sysadmin has been searching in the sources of the MySQL client and came
up with this:

The errormessage (4) is probably (only place where this error could be
found) raised by this fragment in libmysql.c:

if ((sock = (my_socket) socket(AF_INET,SOCK_STREAM,0)) == SOCKET_ERROR)
{
net- >last_errno=CR_IPSOCK_ERROR;
sprintf(net- >last_error,ER(net- >last_errno),socket_errno);
goto error;
}

According to:
http://www-users.cs.umn.edu/~bentlema/unix/syscalls_and_ipc.html
this is what happens:

1) The MySQL client tries to create a new socket.
2) At that moment the thread is interupted by a signal. Control is passed to
the signal handler.
3) After the signal handler has done his work, control goes back to the
original thread. The socket() function notices that it was interupted and
returns the error-code.

If you look at the sample code in section 8.1 of the url above

for (;;) {
rmask = mask;
nfound = select(FD_SETSIZE, &rmask, (fd_set *)0, (fd_set *)0,
&timeout);
if (nfound < 0) {
if (errno == EINTR) {
printf( "interrupted system call\n ");
continue;
}
/* something is very wrong! */
perror( "select ");
exit(1);
}
(...)

The client should try again if the EINTR error was returned.

Did we do something wrong in the configuration of the server or is this a
tiny bug?

Regards, Jigal.

--
MySQL General Mailing List
For list archives: http://lists.mysql.com/mysql
To unsubscribe: http://lists.mysql.com/mysql?unsub=mysql @(protected)