New device status remains unknown.

[bodazyphir]

Hi,

I am a very experienced cacti user and have encountered a problem that seems very common , but I have not yet been able to find a solution anywhere. The problem is only with newly added devices , all existing devices work perfectly. When adding a new device the status of the device never becomes "up" but only remains "unknown". There is also no rrd's created for the device and the device is not added to the poller cache , which kinda makes sense. I can manually add the rrd's but that does not get it added to the poller cache and so the graphs are never populated.

I have really been around the houses on this one , checking rrdtool , php , MySQL , apache , umask settings. The worst part is I cannot even get it to generate a decent error message I can work with.

This is a very pressing problem at the minute .....as we use cacti extensively.

Your help would be greatly appreciated.
Bodazyphir

[cigamit]

First off, what availability options do you have set, and what status do you get at the top of the devices page?

[bodazyphir]

I have tried all the options and they all work (bar TCP ping)

I get a fully system Mib walk and the top and also get an ICMP ping response when this is configured.

It has no effect on the status of the device :-/

[BSOD2600]

Do you have the device set to graph anything? Is it successful? consult the cacti.log -- you might need to increase the logging level to more closely examine for any error(s).

[bodazyphir]

When creating a new device , the mechanism that creates the rra/<hostid>directory in apache works fine .... It does not however create the rrd files , that mechanism seems not to work. I might be due to the host status being "unknown" , but the rrd files are never created. I can manually create them , but due to the device not being populated in the poller cache , the manually created rrd files never get updated. I have refreshed the poller cache , with no luck ....

I have set the logging to debug and that is the most frustrating part , it does not log a single error. That is probly the most vexing part of this issue , is that there is no errors to work with , it's simply decided that when you add new devices it will not create the rrd's and will keep the device in a unknown state. The logging is mainly poller related however , not really for logging issues on new device creation.

I have been trawling the forums and see that this is an issue that pops up periodically since 2004 , but there does not seem to be a definitive fix for it yet. That is why I decided to engage you guys personally , in the hope you can point me in a direction I have not already gone down.

[bodazyphir]

It should also be noted that the device IS being created in the DB , it's just not creating the rrd's. And also not marking the box as "up" and putting it in the poller cache.

09/23/2014 10:25:40 AM - CMDPHP: Poller[0] DEVEL: SQL Exec: "INSERT INTO poller_item (local_data_id, host_id, action,hostname, snmp_community, snmp_version, snmp_timeout, snmp_username, snmp_password, snmp_auth_protocol, snmp_priv_passphrase, snmp_priv_protocol, snmp_context, snmp_port, rrd_name, rrd_path, rrd_num, rrd_step, rrd_next_step, arg1, arg2, arg3) VALUES (151761, 6197, 0,'vruatdma001.iggroup.local', 'se1public', '2', '3000', '', '', '', '', '', '', '161', 'ucd_load15min', '/var/www/html/cacti/rra/6197/151761.rrd', '1', '300', '180', '.1.3.6.1.4.1.2021.10.1.3.3', '', '')"

[cacti@XXXXXXXX log]$ ls -al '/var/www/html/cacti/rra/6197/151761.rrd'
ls: /var/www/html/cacti/rra/6197/151761.rrd: No such file or directory

[BSOD2600]

What cacti version / patches?

[bodazyphir]

0.8.7g , latest 0.8.7g patches

[bodazyphir]

Well it seems I have found the solution. I upgraded to the latest version , no luck. Then I thought perhaps I am just using cacti too heavily as we have +-1400 devices in it with each having multiple data points. So I deleted all non essential devices and that brought the total to 1011 and then it suddenly sprang into life.

So it seems there is a limit to the amount of devices that can be managed , no matter what the hardware platform.

Thanks anyway ..

[BSOD2600]

So it seems there is a limit to the amount of devices that can be managed , no matter what the hardware platform.

there are installs which monitor over 500K data sources using Boost, SSDs, etc. http://forums.cacti.net/viewtopic.php?f ... &start=330 has lots of examples.