06-17-2013 09:23 AM
I'm trying to change the polling frequency of some of the collectors. I'm able to go to Settings > Setup > Collector Status and see all the property groups, their frequency and threshold. However, I have not been able to find where can these values be modified.
Does anyone knows if perhaps this values are default per vendors?
Thanks a lot,
Info: netMRI - IB1102A | Network Automation Version: 126.96.36.199 | Api Version 2.7
05-22-2013 11:32 AM
Good luck with that. I had a support ticket opened requesting the same thing, but received the following:
Currently, we do not have an option to increase the interval or change the behavior.
Please attach 4-6 hrs of SNMP debug logs from the two device for which you have attached the Discovery Diag. So that we can validate what values are return during SNMP polling.
To enable SNMP debug log and gather the file, please follow the instructions below:-
- Device Viewer > Settings & Status > General Settings > 'SNMP Debug' > Enabled > update.
- Once SNMP debug is enabled on the device for 6 - 8 hrs, please have SNMP debug logs collected from DV > Settings & Status > Logs.
- After SNMP debug log is collected, make sure that SNMP debug is disabled.
05-22-2013 11:39 AM
Ha!... okay no thanks, wonder how they will determine if a frequency needs to be change based on 6 hrs of SNMP debug messages.
The only thing I wanted was not to query some devices that frequently.
Anyhow, thanks for your reply.
05-30-2013 03:33 AM
I'm not sure if this is exactly what you want but if you go to Settings > Collectors and Groups and then choose the "Groups" tab you should be able to click the little cog to go to the settings (i.e. Edit Device Group). Then choose the Switch Port Management tab, where you can override the global polling interval and configure your own interval for that group.
05-30-2013 07:30 AM
I did look there, but I only found the where to change the Global Setting. When I edit an individual group I don't see this tab. There are also some other settings that seems to be just global settings and not able to configured individually.
I'm on Network Automation 6.6 and API 2.7
05-31-2013 12:28 PM
The SPM polling interval only affects collection of interface statistics. All of the other data categories, as listed in Setup -> Collector Status, can not be changed. Those are tightly tied into the issue analysis scheduling.
But if you're trying to reduce the polling load on a device with lots of interfaces/ports, it's the interface counters that typically involve the largest table Gets.
02-13-2014 05:39 AM
That is ridiculous! ! !
Our NetMri is collecting config's from switches every hour, and the logs are filled with traces of that.
I can not bellieve, that you can not adjust the interval (like once a month) that the NetMri collects config???
02-13-2014 05:52 AM
SPM polling interval affects polling intervals for switch ports, VLANs, and forwarding tables; if you enable the "collect ARP with SPM data" advanced setting in 6.8 or later, it also affects ARP table collection frequency on switch-routers.
02-13-2014 08:53 AM
Here's my understanding of how config collection works. If I get any of this wrong, hopefully an Infoblox person will correct me.
For Cisco IOS devices, NetMRI polls the ccmHistory table at approximately every 90 minutes. There are three timestamps that define when the running-config was last changed, the startup config was last changed, and when the running-config was last saved. NetMRI uses these to decide if a configuration change has happened since the last collection was performed. If so, then the config collector script is launched to collect the current running and startup configs.
The new running and startup configs are compared to the previously stored revisions. If there are differences, the new ones are stored; if not, they are not stored but the "last collected" timestamp is updated (top of the Config Explorer listing). If the timestamps indicate that the running-config has not been saved, that device will be cited in the Config Running Not Saved issue.
So what could cause NetMRI to continue to login and collect the configs every 90 minutes, when the differences comparison shows no actual changes? If one enters and exits config mode in an IOS device, yet makes no actual config change, IOS will still produce a syslog CONFIG message indicating that a change was performed. And of course, the ccmHistory running-changed timestamp will be updated. You can complain to Cisco for never making a smarter assessment! Subsequently, when NetMRI polls the table, it believes that the running-config has been changed, performs config collection, compares the results, and does not store any new revisions.
Could NetMRI be smarter about this? Yes, since it has the previous three values, on the *next* poll it could notice that they are unchanged, and not perform a needless collection. [Infoblox: if that enhancement was ever added, please say so.]
What can a customer do to prevent this behavior? When someone exits config mode without making any change, they can issue a "write memory" so that the running-saved timestamp is later than the running-changed value. That will prevent the device from appearing to have unsaved changes. In NetMRI, in the Config Running Not Saved issue list, you can select all of the devices and schedule a job that will perform the save.
Note: besides polling, NetMRI can be asynchronously notified of a config change. Configure the NetMRI as a syslog destination with "logging host x.x.x.x". NetMRI will receive all syslog messages but will ignore all except the CONFIG type. Upon receipt, an immediate collection will be launched. Another really good reason to use this method is that NetMRI will store the message content that shows the username and location of the changer, which is essential for auditing.
Note: if you still have any old CatOS-based switches, those do not support the ccmHistory table OID. NetMRI will login and collect the configs every 90 minutes because there is no other way to identify if changes have been performed.