0.8.8b ignores database settings in include/config.php

[Wavebourn]

Hi!
I've installed the latest 0.8.8b version on the machine that was running 0.8.7g, transferred database settings as is:

$database_type = "mysql";
$database_default = "cacti";
$database_hostname = "127.0.01";
$database_username = "cactiuser";
$database_password = "xxxxx";
$database_port = "3306";
$database_ssl = false;

but it returns:

FATAL: Cannot connect to MySQL server on 'localhost'. Please make sure you have specified a valid MySQL database name in 'include/config.php'

No matter what host name I set for $database_hostname it always returns the same, "Can not connect to MySQL server 'localhost'"

When I telnet to the port 3306 no matter on localhost or fully qualified host name, I see response of the server, and version 5.1.61

How can I debug it?

Anatoliy

[Hitch]

I've run into this problem while installing 8.8 on a CentOs machine, and as it turns out, I couldn't use 127.0.0.1 for the hostname, I had to use the name I gave the machine, and lo and behold, it worked!
Mine ended up looking like this:

$database_type = "mysql";
$database_default = "cacti";
$database_hostname = "blogunit2b";
$database_username = "cactiuser";
$database_password = "xxxxx";
$database_port = "3306";
$database_ssl = false;

I have no idea why that did it, but give it a shot, maybe it will work for you too.
Or try "localhost"


Roger

[Wavebourn]

I believe database after multiple upgrades had such data so it was making php scripts braindamaged. Actually, 0.8.8a worked well while I was deleting very old hosts and graphs that did not exist anymore, but after removal of the device that was on top of the list this things started.
Version 0.8.7g kind of worked offering to "upgrade" from 0.8.8a to 0.8.7g then complaining that such upgrade is not possible. Version 0.8.8a prompted for login/password, authenticated, showed starting page, after that any click that requires acces to database resulted in "Can not connect..." message. After restarting mysql it worked again. Version 0.8.8b even did not ask for password, it was already brain-damaged from the start.

I decided to reinstall everything and configure everything from scratch. New life with new look!

All settings are the same, but it works fine. Looks like some buffers were overflown by data.

[Hitch]

Well, good to hear you got it working. I 'upgraded' from 8.7a from 2 machines both running redhat ES3 on 2 separate networks, and were used very heavily, device wise and user wise, over 200 devices each and I can't even recall how many ten's of thousands of data sources. I was using thold, and the baseline deviation worked well, but I could only run 1 poller process, as thold broke the system if I ran more than one process.
After upgrading to 8.8a, on a new machine, I can run multiple process's with thold, however baseline deviation for thold is completely messed up in the new build (4.9 I believe), and I cannot figure out how to make it give accurate results, which is a real disappointment, but at least it emails me, so I guess thats a plus even though the info it sends is worthless.

Roger