graph do not appear

[derrick.joseph]

I was able to install cacti-0.8.6c.fc3.i386.rpm on a Fedora Core release 3 (Heidelberg) without any problem automatically creating the cacti directory at /var/www/html.

Since I installed the rpm as root, the default user and group of the cacti directory is being owned by the user root. I added a user on the system named "cactiuser" and changed the ownership of rra/ and log/ to cactiuser leaving other files and directories owned by root and under the group root. The setting on my /etc/crontab is:

*/5 * * * * cactiuser /usr/bin/php /var/www/html/cacti/cmd.php > /dev/null 2>&1

where cactiuser, was the user I added in the system.

After successfully accessing http://localhost/cacti, I created a new device without any problem. Everything works as I can see the status, current (ms), average (ms) and availability working. The problem is, I can't see any graph ... I checked on my /var/www/html/cacti/rra directory and nothing is in there. My polling is also working as my cacti.log logs the actions such as:

02/11/2005 07:30:02 PM - CMDPHP: Poller[0] Time: 0.6887 s, Theads: N/A, Hosts: 1
02/11/2005 07:35:02 PM - CMDPHP: Poller[0] Time: 0.5879 s, Theads: N/A, Hosts: 1
02/11/2005 07:40:01 PM - CMDPHP: Poller[0] Time: 0.5950 s, Theads: N/A, Hosts: 2
02/11/2005 07:50:02 PM - CMDPHP: Poller[0] Time: 0.6767 s, Theads: N/A, Hosts: 2
02/11/2005 07:55:02 PM - CMDPHP: Poller[0] Time: 0.6220 s, Theads: N/A, Hosts: 2
02/11/2005 08:00:02 PM - CMDPHP: Poller[0] Time: 0.8318 s, Theads: N/A, Hosts: 2
02/11/2005 08:01:36 PM - CMDPHP: Poller[0] Time: 0.3888 s, Theads: N/A, Hosts: 2
02/11/2005 08:05:01 PM - CMDPHP: Poller[0] Time: 0.6421 s, Theads: N/A, Hosts: 2
02/11/2005 08:10:02 PM - CMDPHP: Poller[0] Time: 0.6359 s, Theads: N/A, Hosts: 2
02/11/2005 08:15:02 PM - CMDPHP: Poller[0] Time: 0.7194 s, Theads: N/A, Hosts: 2

What seems to be the problem in here ? Please help as I am very eager to use it.

Derrick

[rstarr]

My best guess would be that you are using ICMP (as opposed to UDP) as your ping type under "settings, poller" It requires it be run as root to use ICMP. I think thats what happened to me at one point down the road. Since my cacti processes aren't running as root, it couldn't use ping, so I got a very quick "no reply" from them since it couldn't ping them at all. I think those small numbers less than one second is the time it takes to attempt to run the ping command on your two hosts. For me, it took around 3 seconds for 16 hosts.

I believe, and someone correct me if I'm wrong... if cacti cannot ping the host and finds it dead, it will not even attempt to grab SNMP info from it. It seems logical to me, but I haven't looked at the code.

I would suggest using UDP for the ping type. Also use "Ping and SNMP" for the downed host detection.

Another _possible_ thing would be that your hosts have a very high latency (400+ ms) and it detects them as being down since it exceeds the default timeout. This probably isn't the case unless you are going through a conjected or slow speed WAN link. If you suspect this, do a ping from your cacti box to get an idea of how high the response is. You might need to tune that on your "Ping timeout value" on "settings, poller".

Hope that helps some,

Let me know what happens.

-Russ

[TheWitness]

Very good information relative to the poller Russ! Both have to succeed in order for the host to pass inspection. Another good method is to turn logging to DEBUG for 1 pass. You will get much better information in DEBUG, but much more as well.

TheWitness