1.0.1 to 1.2.3 - good except for a few graphs wont draw

[camerabob]

Check for differences between the rrd directories and files. There may be rights issues hidden or masked there.

[kjstech]

Check for differences between the rrd directories and files. There may be rights issues hidden or masked there.

Ok here is the rra directory permissions under /var/www/cacti

Code: Select all

drwxr-xr-x 3 root     root     233472 May 14 10:30 rra

Core switch that only works with CMDPHP

Code: Select all

-rw-r--r-- 1 root root 9006984 May 14 12:29 yellowcore_traffic_in_4088.rrd

Core switch that only graphs with SPINE

Code: Select all

-rw-r--r-- 1 root root 9006984 May 16 16:22 purplecore_traffic_in_2535.rrd

Same permissions. Just one only updates with cmdphp and one only updates with spine. Can't figure this one out.

[netniV]

The likelihood is that you are running the cron as a specific user but have spine with the setuid so it can be root automatically. The rra folder should always be read/write to the website and polling users.

[kjstech]

Yes, cron runs cmdphp as www-data it looks like, not root. I’ll be back in the office Monday to play with permissions some more.

[kjstech]

I changed /var/www/cacti/rra ownership recursivly to www-data:www-data and also made it wide open 777 permissions.

Still, spine will graph all but our one core switch. cmdphp will graph that one core switch but will not graph a handful of other devices. The RRA's for these different devices have identical permissions:
-rwxrwxrwx 1 www-data www-data

in visudo I added this line to make sure www-data was running the poller. I don't know, maybe this is old advice.
root ALL=(www-data:www-data) /usr/bin/php /var/www/cacti/poller.php

[kjstech]

I never could figure this out so I created a new device for this core switch that does not graph with spine.

I used the OSPF Lo0 IP address for it (the core switch has multiple interfaces obviously, at the core). I changed SNMP timeout from the default 500 to 1000. It now is getting data.

On the original device I also tried changing SNMP timeout from 500 to 1000 and that didn't make a difference I'll end up deleting that device since the new one works, it just has -OSPF at the end of the device name since I'm using that alternate IP interface.

[Osiris]

There are some additional fixes to the indexing functions in 1.2.5. Should make things a bit more stable I think.