I know version 5.1.x has been released but there is still much debate on wether this is safe enough to use in a production environment. I haven’t fully tested it yet in all the different environments and setups I would like to (mostly due to time constraints) before I commit to this new release.
This (5.0.75) is the version that comes after 5.0.67. This is a bug and security fix release and it is available in source format only. Because this is a security fix I would urge you all to upgrade your MySQL server installation.
Check all the fixes that are listed on the release notes to see what issues are resolved.
I’ve compiled this version and did some simple tests on Leopard and Tiger and both can be compiled and installed using the instructions in the documentation set without problems
Comments
5:29 am
First, Happy New Year!! and thank you for all your help getting me and all the rest of us up and running on Mac-based servers!
I just upgraded to MySQL 5.0.75 and I’m now getting repeated warnings in mail.log:
Dec 31 19:58:52 s postfix/trivial-rewrite[3380]: warning: connect to mysql server localhost: Can’t connect to local MySQL server through socket ‘/tmp’ (38)
These started as soon as I updated the mysql server (shortly after which I restared the whole computer, so I know that all my processes were relaunched).
/tmp/mysql.sock exists and has the right permissions.
I’m at a loss…any suggestions?
Thanks!
6:40 am
Adding one bit to my previous comment, which I think holds some clue to the issue, but I am also seeing this message:
Dec 31 21:33:12 s authdaemond[95]: authmysqllib: connected. Versions: header 50051, client 50075, server 50075
My conclusion is that the “Header” – whatever that is – is causing this confusion.
courier-auth connects to mysql fine, as do most of the postfix processed (all of my alias and virtual tables are in mysql. So far, I’ve only seen canonical_maps fail.
PHPMyAdmin also connects via the socket with no issue but reports the server version as 5.0.75 but the client version as 5.0.51b.
I think the clue is there…but I just don’t know…
2:00 pm
Problem solved, somehow Jeff made a mistake during the configuration fase and set the socket to be in /tmp instead of the full /tmp/mysql.sock.