Bug #19437 Connection refused by server: "2002 Can't connect ... /master.sock (61)"
Submitted: 29 Apr 2006 16:37 Modified: 4 Jul 2006 14:00
Reporter: Joerg Bruehe Email Updates:
Status: Closed Impact on me:
Category:MySQL Server Severity:S2 (Serious)
Version:5.0.21 and others OS:MacOS (Mac OS X (all CPUs))
Assigned to: Mats Kindahl CPU Architecture:Any

[29 Apr 2006 16:37] Joerg Bruehe
rpl000002                      [ fail ]

Errors are (from PATH/mysqltest-time) :
mysqltest: Could not open connection 'default': 2002 Can't connect to local MySQL server through socket 'PATH/master.sock' (61)
(the last lines may be the most important ones)

Killing Possible Leftover Processes
Happens on all OS X machines ("powermacg5", "osx-imac-x86"; "xserve-a" both 32 + 64) in various products / tests, 32 failures in total in the 5.0.21 build.
Errno 61 is "ECONNREFUSED" (checked on "osx-imac-x86").

Less often: same symptom, same platforms, in test "kill"
(which more often shows other failure symptoms, reported as bug#16850, bug#16851, and bug#19291).

The combination of "Can't connect ..." through "master.sock" with error 61 is specific to OS X, but it seems non-deterministic.

How to repeat:
Run the test suite on OS X.
[16 May 2006 18:10] Chad MILLER
I'm stuck on this bug.  I'm attaching my most recent slave.trace .
[16 May 2006 18:13] Chad MILLER
one thread frees structure before the other uses it; grep for "0x4304a18"

Attachment: 19437-slave.trace (application/octet-stream, text), 195.66 KiB.

[16 May 2006 19:39] Chad MILLER
last 5000 lines of slave.trace; grep for "0x43047e8"

Attachment: slave.trace (application/octet-stream, text), 194.81 KiB.

[26 May 2006 14:40] Elliot Murphy
I believe this is also showing up intermittently in pushbuild on SLES9-x86. Crashes during slave shutdown.
[20 Jun 2006 18:47] Bugs System
A patch for this bug has been committed. After review, it may
be pushed to the relevant source trees for release in the next
version. You can access the patch from:

[26 Jun 2006 4:22] Elliot Murphy
Noticed fix was pushed from team tree to main tree for 5.0.23.
Merged to 5.1 tree for 5.1.12.
[4 Jul 2006 9:33] Mats Kindahl
The problem occured sporadically when executing the replication tests on an OS X machine. After running all replication tests ~280 times with no error, it seems as if the bug is fixed. There are no specific documentation needs for the bug since it does not change any behaviour.
[4 Jul 2006 14:00] Paul DuBois
Per Mats, no changelog entry is needed.