Kaseya Community

Agents Falling Offline - Cannot Refresh.

  • I am finding certain agents keep falling offline. They are on networks with other agents that are staying online just fine. When I have a local user right-click refresh the agent from the system tray, message = "the agent is already trying to connect to the Kserver, this refresh request will run as soon as the current connection completes."

    Rebooting will resolve temporarily, but the next morning agent will be offline. Agent/System/Application/Script logs have no common denominators between the machines that are having this problem. These logs seem fine.

    It is only a handful impacted, but I need to get the issue resolved.

    Note these problems began after upgrading Kaseya and the agents to the latest this last week.

    Brian


    Legacy Forum Name: Agents Falling Offline - Cannot Refresh.,
    Legacy Posted By Username: bgilmore
  • I've had the problem on a few specific machines all along, so for me at least it's not an upgrade issue.

    Legacy Forum Name: Agents,
    Legacy Posted By Username: tom@tbkconsulting.biz
  • more info:

    upon rebooting the machine, the agent comes online and the log then populates with these entries for the enitre time the agent was offline:



    "10:53:04 pm 13-Feb-07 ERROR: keep-alive in doClientTasks() call to kascSend failed with error -12
    10:52:04 pm 13-Feb-07 ERROR: keep-alive in doClientTasks() call to kascSend failed with error -12
    10:51:04 pm 13-Feb-07 ERROR: keep-alive in doClientTasks() call to kascSend failed with error -12
    10:50:04 pm 13-Feb-07 ERROR: keep-alive in doClientTasks() call to kascSend failed with error -12
    10:49:04 pm 13-Feb-07 ERROR: keep-alive in doClientTasks() call to kascSend failed with error -12 "




    Legacy Forum Name: Agents,
    Legacy Posted By Username: bgilmore
  • Are you running any scripts on those machines which you're not running on other machines?

    Legacy Forum Name: Agents,
    Legacy Posted By Username: bpenland
  • no they are all the same and upon checking the script log, no scripts seem to be running at about the time they go down. Still, what was your thought about the scripts?

    Legacy Forum Name: Agents,
    Legacy Posted By Username: bgilmore
  • If the agents are always going offline at the same time each morning and random agents are affected it might be due to some resource intensive activity on the server side (e.g. all agents running audit at 9am). You might want to set up some performance counters on the server (in particular I/O counters where your SQL database sits such as disk queue length and % disk read/write per sec). Also check your available network bandwidth.

    Just some thoughts...

    -Ed


    Legacy Forum Name: Agents,
    Legacy Posted By Username: bellcpa
  • have you made sure there isn't some limitation in how many active connections you can have out to the internet in your router/firewall? The new agent is much more agressive with connections.

    For example under the old agent for 25 machinesa firewall might count 15-20 connections because all the agents werent checking in at the same time. Under the new agent, all 25 would be active a the same time.


    Legacy Forum Name: Agents,
    Legacy Posted By Username: far182
  • Very good point, I would check the firewall first Wink

    Legacy Forum Name: Agents,
    Legacy Posted By Username: bellcpa
  • firewall on the stations has been checked. Further more, if it were firewall it would be all or nothing, right? The clients work for quite sometime, even a day or two after reboot. Then they fall off. Also if it were the firewall on the network, why some machines and not others. Still we have checked the PIX unit to verify if maybe there is a cap on connections, there is not. And even so strange it would typically be the same machines, plus a new one now and again.

    Will implement the Kserver monitors though per the suggestion below and update upon seeing results.


    Legacy Forum Name: Agents,
    Legacy Posted By Username: bgilmore
  • bgilmore wrote:
    firewall on the stations has been checked. Further more, if it were firewall it would be all or nothing, right? The clients work for quite sometime, even a day or two after reboot. Then they fall off. Also if it were the firewall on the network, why some machines and not others. Still we have checked the PIX unit to verify if maybe there is a cap on connections, there is not. And even so strange it would typically be the same machines, plus a new one now and again.

    Will implement the Kserver monitors though per the suggestion below and update upon seeing results.


    When you said PIX I felt I should comment. I love PIX, great product. And yes... if it was a FILTER problem it should be all or nothing. Workstation firewall the same.

    What I was talking about is connections. How many active connections your firewall - and if seperate Router allows. Many PIX have this limitation.


    Legacy Forum Name: Agents,
    Legacy Posted By Username: far182
  • Thank you for your help. As I mentioned we checked to see if there might be a cap on the active connections on PIX. No cap. (or at least not one that should impact our enviroment. I will double check with the admin who manages this unit, but he did tell me that he specifically looked for this possibility.






    Legacy Forum Name: Agents,
    Legacy Posted By Username: bgilmore
  • I just ran into a similar situation (not with agents, but a machine falling off the network).

    There was a PIX 501 (before us). After checking, they had the 10 user version. They were running 11 machines.

    PIX's come in in 10, 50 and unlimited version.


    Legacy Forum Name: Agents,
    Legacy Posted By Username: rfouche
  • Still remember that in my situation, I can get ALL machines up at once. I just need to reboot the offending machine that fell off. Then all clients are active at one time until the next check in the morning. (they drop typically at night). If there was a problem with the number of active connection limit, then it would continue to impact regardless, correct? The norm for our system is to have all PC's up and running all the time. They normally are never shutdown unless we do so on purpose.

    Legacy Forum Name: Agents,
    Legacy Posted By Username: bgilmore
  • I think the biggest clue in this mystery is this:

    "When I have a local user right-click refresh the agent from the system tray, message = "the agent is already trying to connect to the Kserver, this refresh request will run as soon as the current connection completes."


    Why would these agents fail to connect as though they are caught in a loop that only rebooting resolves?




    Legacy Forum Name: Agents,
    Legacy Posted By Username: bgilmore
  • Is this happening at all of your customer sites?

    Also, check the KServer to see how many connections it is maintaining (netstat -a from the command prompt on the KServer).

    It sounds like the agent is making a connection to the KServer with a specific source port, and then is moving to a different source port.

    Have your PIX admin setup a capture on the firewall between a known bad agent and the KServer. Have him capture the raw data to see how the traffic is being handled.

    The message you are getting really means that the connection was initated but it has not had a response from the KServer.


    Legacy Forum Name: Agents,
    Legacy Posted By Username: bpenland