Server Manager Monitor Notification: Outstanding Requests

navin_JDE

Active Member
Hi List,

We were getting continous SM alerts with the following messages to our mails as shown below..

<font color="green"> </font>
You are receiving this message because a monitored event has occurred. Details of the event are shown below.

Monitor: Production Monitor
Event Type: Outstanding Requests
Event Time: 9/13/09 3:36 AM
Managed Instance: HGN01A109
Additional Information:
Enterprise Server Process Detail
======================================
OS Status: 1 (RUNNING)
E1 Status: 1 (BUSY WITH STHG)
AS/400 Job Id: 0
Process Id: 5128
User Id: 0
Group Id: 0
Process Type: 0 [Kernel Process]
Start Time: 1252794790000
Last Updated: 1252794790000
Process Name: XMLPUBLISHER KERNEL
<font color="green"> </font>

Did anyone ever faced this... how to resolve this.. we were getting this only for our production servers and for some kernels like XMLPUBLISHER KERNEL, XML DISPATCH KERNEL, EVN KERNEL and XML List Kernel.

Thanks in advance.

Naveen Mortha,
E1 8.11sp1, 8.98.0.4, Windows, SQL 2000
 
First, let's understand outstanding requests. When a message is sent to an enterprise server it is dispatches to the appropriate kernel based on the type [call object, xml publisher, etc]. The network process puts that request into a queue in shared memory. The kernel process reads these messages from the queue, processes them, and puts them back in a queue for the network process to send the response back.

Nearly all aspects of these queues are controllable -- refer to the server manager documentation for the related INI settings. That said, be careful when making changes to these.

So, what about your post. Well, it may or may not be a problem. Processing time for messages takes a variable amount of time. Security server requests should be near instantaneous. CallObject requests take as long as the called business function takes. XML publisher requests take as long as the they take. Helpful, huh?

My point is that queued messages may or may not be a problem. Are your users complaining [reasonably of course] of long processing times? If not you may simply have your threshold for outstanding requests monitor in server manager set to a too low value, and thus you are being warned about something innocuous.

Another thing to note about the outstanding requests monitor: server manager keeps track of the last value sent out, and only sends a new alert if the current condition exceeds that. I am not sure if that is a global value, by kernel type, or by actual kernel process id. I suspect it is the later, but am not sure.

If you get all the sudden getting many of these alerts for the same process it may indicate the process has hung -- that is it has stopped processing incoming requests, new requests are queuing, and the clients will start getting timeouts.

Bonus note on timeouts: there are several places this can occur. The client can time out because its defined maximum threshold for waiting for a response has elapsed, or the server may send back a message saying the maximum time a message is allowed to be pending in a queue has occurred. Either way it looks like poop to the client, and either way its time to investigate why the kernel process is taking so long to process messages or has stopped processing them altogether. Time to dig into those logs.
 
Hi,
Thanks for ur reply. By reading this i got some idea abt this outstanding requests. By going through logs we found the user PSFT is locked after resetting the user and restarting the services on Entserver resolved the issue.

Thanks,
Naveen
 
Back
Top