Autodiscovery performance

Hello Didier,

It will only be possible to run the auto-discoveries at a particular time. We recommend to run the discoveries when PRTG is not actively used by colleagues. This will keep the load at a low level. You can change this time on the Setup > System Administration > Monitoring page.

Best regards, Felix

Hello,

Am I correct to understand that there is no option in limiting simultaneous auto discoveries ? If we want to add a complete subnet in one auto-discovery group it scans / autodiscovers about 10 ip addresses at the same time. I would like to be able to limit that amount.

thank you for your feedback!

Didier

Hello Didier,

The number of concurrent auto-discoveries is restricted to 10, I'm afraid that this value cannot be changed.

Best regards, Felix

Hello Felix,

Thank you for your feedback. Will this change in future versions of the product ? We are very happy with it but this something that is really impacting adding devices and sensors.

Didier

Hello Didier,

I'm afraid that there are no current plans to restrict the number of concurrent discoveries in the foreseeable future, sorry Didier.

Best regards, Felix

I have the same issue .. as soon 10 discovery run on a subnet .. it kill the server performance.

any way to change the number of auto discovery at 5 .. i'm ok ..

Hi,

The number is still fix, please bear with us.

Best regards, Felix

*FEATURE REQUEST PLEASE!!!* Please mark myself, and if a weighting is required, the 50+ clients who also exist on our two hosted systems (XL1 licenses on both) as requesting this be a configuration that we can make a change to. Even if it means having to perform a registry edit, I'm game.

We have several Unlimited/XL1 licenses (two which we host and at least 2 that are run by clients) where the performance of the Core system is terrible, the application essentially turns unusable for well over an hour, whenever auto-discovery is run. With this many clients, across 5+ timezones, it is unacceptable for us to say "Well geez guys, maybe you should limit your auto-discovery options to when other users are not logged in using the system" - that just is not possible.

It;s not like we are running skimpy hardware either. While each of the servers was bought/specc'ed at different times, each is at least an Intel Xeon, 4 core CPU with minimum 16 (usually 32 and sometimes 64) GB of RAM. We have one of the systems on 10K SAS drives, another with 15K SAS drives with a PERC7 RAID 2GB Adapter, another with Enterprise-class Solid State Drives, and even the Solid State Drive system runs slow as molasses when doing auto-discovery with 10 simultaneous threads. Sensor counts aren't unreasonable on any of these (9000 sensors on the Solid State Machine, 13,000 sensors on the 15K SAS Drives).

We really need a solution as some of our customers have started to leave us while others are threatening to leave our service and the situation doesn't look good.

Thanks for the consideration! Todd

• FEATURE REQUEST PLEASE!!!* We are considering to use different monitoring tools due this problem with auto-discover. When auto-discover run over more than 3-5 devices there is a terrible performance issue, we have observed also how we have lost also some monitoring.

Right now only solution it's to do auto-discover 1 by 1 by hands. After many years without a solution we are thinking seriously to move to another monitoring tools.

Hi ZFS_IEIEN_Network,

Which version of PRTG are you running in the installation where the auto-discovery causes performance issues? To get a better overview of your hardware and configuration, kindly forward a support bundle by using the Setup > Contact Support button of your PRTG installation. We'll then be able to see if we can give you any advices.

Best regards, Felix

Hello

I would like to add to this discussion, in a different direction: Sometimes, when adding new devices or running 3am maintenance, responsiveness is a very low priority, and it would be useful to be able to increase the maximum number of autodiscoveries running at once.

Thanks!

Thanks for your input Ian! The performance of the auto-discovery got significantly improved over the last years. How many scheduled auto-discoveries do you run and how many hosts are scanned so that you feel like that the number of simultaneously discoveries should be increased?

Best regards, Felix

Hello, when adding the devices of new remote shop we have the autodiscovery with switch template of 4 devices. PRTG (VM core) 10k sensors with 8 CPU and 32 GB of RAM takes approximately 4 minutes to complete 1 autodiscovery of 1 switch during this process the resources of the machine are OK not overloaded but the circa 2000 ping sensors reported "Unknown status" even though server resources are free. Is it possible to limit concurrent autodiscovery operations to a single operation only? Thanks, Regards.

Hello Emiliano,

auto-discovery cannot be configured in this way. Usually it should work, however on larger systems like with 10k sensors, it could impact performance. Even if there are resources free, because the bottle neck is writing to the device tree. At that time, other threads cannot access the tree. It could be that tasks pile up, so that the core service stalls for some time.

For in-depth analysis please contact us through PRTG Setup / Contact Support, and include the support bundle.