Incidentally, I finally resolved the
server problem – as Peter suggested, it was just a matter of reconnecting
to the MySQL database on occasion. Not sure why switching from Fedore Core 3
to FC 5 required that change, however.
Thanks for the suggestion Peter, and sorry
to everyone for the inconvenience!
-david
From:
Sent: Tuesday, October 24, 2006
11:23 AM
To:
Subject: Re: [iglance] MySQL 5
oddities?
This way maybe you will also find MySQL
replication and failover a fun.
----- Original Message -----
From: David Barrett
Sent: Monday, October 23, 2006 5:26 PM
Subject: RE: [iglance] MySQL 5 oddities?
Ah, good idea. I’ll check that, thanks.
From: iglance@yahoogroups
.com [mailto:iglance@ ] On Behalf Of Peter Panyahoogroups. com
Sent: Monday, October 23, 2006 4:08 PM
To: iglance@yahoogroups.com
Subject: Re: [iglance] MySQL 5 oddities?
IMO, it is harmless to retry/reconnect/ failover when an error such like
CR_SERVER_LOST CR_SERVER_GONE_
ERROR CR_SERVER_HANDSHAKE
_ERR
occurred to mysql_store_ result().
Peter ----- Original Message -----
From: David Barrett
Sent: Monday, October 23, 2006 1:41 PM
Subject: [iglance] MySQL 5 oddities?
So the iGlance server suffered a motherboard failure, thus triggering a big upgrade of hardware and software alike. One change to the system is now we’re using MySQL 5, and something strange is afoot.
Basically, the connection between the iGlance server and MySQL just stops working on occasion. No other changes to the code, and it works great for hours or even days, and then spontaneously stops working. The server in this state is still “working” fine in that it is processing and responding messages, but anything involving the database fails to work. Restart the server without touching MySQL at all, and it works fine for another day or so.
Any ideas? Is there some kind of age or query limit on a connection, thus requiring an occasional reconnect?
-david