[solved] Gaps in graph more frequent as devices are added

johnoliveira · Post by **johnoliveira** » Mon Feb 09, 2009 7:35 pm

I am running Cacti (Nated) on W2003S in a VM enviroment behind a cisco firewall. Currently we poll less than a 1000 hosts but growing by the day. I forsee 5000 plus hosts very near future.

The graphs have random breaks in data. It may be one to 5 polls missing at a time in sequence randomly thru the hosts.

We graphed the local host cpu, mem etc and they all look good. Nothing seems to be taxed. We have try both pollers and have had the best results with cmd. My specs are below. Could the fix be a simple as adding memory (limit)?

I have provided a sample graph ..

Cacti Version 0.8.7b
Cacti OS win32
SNMP Version net-snmp
RRDTool Version RRDTool 1.2.x
Hosts 830
Graphs 7017
Data Sources Script/Command: 26
SNMP: 728
SNMP Query: 6332
Script Query: 2
Script - Script Server (PHP): 256
Script Query - Script Server: 4
Total: 7348
Poller Information
Interval 300
Type cmd.php
Items Action[0]: 13282
Action[1]: 30
Action[2]: 264
Total: 13576
Concurrent Processes 8
Max Threads 15
PHP Servers 1
Script Timeout 25
Max OID 10
Last Run Statistics Time:194.4903 Method:cmd.php Processes:8 Threads:N/A Hosts:831 HostsPerProcess:104 DataSources:13576 RRDsProcessed:4749
PHP Version 5.2.5
PHP OS WINNT
PHP uname Windows NT xxxxxxxxxxxxxx 5.2 build 3790
PHP SNMP Installed
max_execution_time 30
memory_limit 128M
It is highly suggested that you alter you php.ini memory_limit to 232M or higher. This suggested memory value is calculated based on the number of data source present and is only to be used as a suggestion, actual values may vary system to system based on requirements.

Post by **TheWitness** » Mon Feb 09, 2009 10:31 pm

A snapshot of your Cacti log file is more relevant to understand the crux of this problem.

TheWitness

BSOD2600 · Post by **BSOD2600** » Tue Feb 10, 2009 1:24 am

Any reason why you haven't raised the php memory_limit per the suggestion?

johnoliveira · Post by **johnoliveira** » Tue Feb 10, 2009 8:38 am

No not really. However we shall give that a whirl.

johnoliveira · Post by **johnoliveira** » Tue Oct 20, 2009 3:08 pm

Followup Problem solved.

Turned out that when I polled the return data amount was much larger than I thought. 100mps switch could not handle 90 meg of data all at once. So went to a gig on gig pipe.

Thanks

drewshin99 · Post by **drewshin99** » Tue Oct 20, 2009 6:42 pm

seems like a awfully high amount of traffic...i have 1300 hosts with 25000 datasources and it uses about 100Kbytes/sec of traffic at most.

johnoliveira · Post by **johnoliveira** » Wed Oct 21, 2009 9:07 am

I am just at 20000 cpe's. Will be adding alot more. Or at least my peeps will be keying them in. Thanks

[solved] Gaps in graph more frequent as devices are added

[solved] Gaps in graph more frequent as devices are added

Who is online