Kaseya Community

4.7 and SNMPGET.exe high usage

  • Hello,

    Since upgrade to the 4.7 we have had 2of customersHP ML350 G4 running Windows 2003 servers start reporting high CPU usuage eventually leading to lockups requring hard reboots several times in a day.

    I noticed that SNMPGET.exe would lauch several instances in taskmgr (15 or so) and CPU would spike to 100% for few seconds and then stop, this was repeating several times within a minute interval.

    This also resulted in overheat issues as well as the servers were running busier than before.

    Removing SNMP community from these servers addressed the issue for now.

    Has anyone noticed SNMP taxing the server load? is this simply a coincendence or did something change in 4.7 that would cause this?


    Legacy Forum Name: 4.7 and SNMPGET.exe high usage,
    Legacy Posted By Username: Farru21
  • Hi,

    a multi-thread type SNMP capability was added in 4.7 so that each element of the SNMP monitor set can be retrieved at the same time. If you have a LOT of elements in your SNMP monitor set it may get busy. (usually most people only ever monitor one or two things on each SNMP device, like through-put or availability)

    The kind ofCPU usage you are talking about here does seem over the top though, and I would suggest you get in touch with Kaseya support to let them know about this by going to System->Request Support inside your Kaseya server and raising a ticket.

    Legacy Forum Name: Technical Issues,
    Legacy Posted By Username: raybarber
  • Hello Ray,

    These are HP Proliant SNMP Sets monitoring FAN, CPU, RAID and other SNMP elements useful to track SNMP health of a HP Proliant server.

    We use these because theKaseya Monitoring via WMIare not able to alert us on these FAN, RAID status etc. We could rely soley on the event log monitoring to alert us however with the possibility of the overflow we did not want to take this to chance.

    I am guessing that we will not able to use this set with 4.7 anymore, what other alternatives would you recommend to monitor hardware health?


    Legacy Forum Name: Technical Issues,
    Legacy Posted By Username: Farru21
  • Not that I want to continue to rant. But I've done several posts on getting Kaseya to revise the overflow logic. Farru21 has just admitted he's avoiding using event log monitoring due to the overflow issue!

    Now to the point:


    I'm not just ranting for my own sanity. I've programmed before and I know why they have the overflow logic. However, just disablingevent log monitoring is a joke.Kaseyayour hanging your partners out to dry!!!!

    Quick example time line:

    11pm - backup starts, 1am - several backup related event entries happen, a overflow occurs, event log monitoring is disabled, 3am - backup job finishes with errors, 5am - system fan fails, 8am - event log monitoring is manually re-enabled, logs have to be check manually for problems. (I'm speaking on the terms of non 24 hour staffed environment)

    In this time line one or more CRITICAL events may have been missed or at minimumthe approiate corrective action was delayed. We also have to take time to manually review the logs. Oh wait, isn't that one of the main reasonswhy we brought a solution to do it for us!?!?!?! We rely on Kaseya to filter and delivery critical event log entriesdirectly to us. However, the currentoverflow logic makes it very unreliable. Even filtered eventscan trigger an overflow!

    Kaseya if you want to make a lot of partners happy, put parameters to automate re-enabling the monitoring. Allow partners to control the overflow logic with warnings stating "Changing these settings may create additional load on the agents and Kaseya servers, please do not change them without reading the help first.". Also adding log trending would be great too. There are often event entries that we don't care about unless there are several in a period in time. Currently there's no way to set a event log monitoing threshold for each event or type.

    Legacy Forum Name: Technical Issues,
    Legacy Posted By Username: connectex