SMS: How Senders Interact w/ Inactive or Congested Sites and Retrying Jobs

ID: Q214485


The information in this article applies to:


SUMMARY

In situations where bandwidth is on a normal basis completely consumed by Systems Management Server (SMS) Sender activity:

If one or more Sites that a Sender is targeted to becomes inactive or unreachable for any reason, other Sites that are functional that are also targeted by the same Sender can be severely impacted causing their Jobs to Fail or go into Retrying. Typically, Senders configured to use RAS over a slow modem or using a congested WAN link is where this problem occurs. The reason for this behavior is when Jobs are targeted at Sites that are inactive or down, more bandwidth is consumed by these attempts at 'Retrying' the down Site. The Job, Scheduler, and Sender 'Retry' attempts to the inactive Site consume more time (bandwidth) retrying than if the original Job had completed successfully (assuming the original job was fairly small). This problem can be exaggerated by limiting the (schedule) availability of a senders outbox.


MORE INFORMATION

Several aspects of this problem are detailed here:

  1. The RAS and SNA Senders will retry a specific scheduled Send request five times and two times for the LAN Sender. The number of retries cannot be set to zero.


  2. The Sender's RAS or LAN will retry for these five times at the interval specified in the following registry location:


    HKey_Local_Machine\Software\Microsoft\SMS\Components\SMS_Async_RAS_Sender...
    "Number of Retries" and "Retry Delay in Minutes"
    Or also set in the Sender Manager as
    Retry # Times every # Minutes
    This is set by default to five Times for every one Minute.

    NOTE: Setting this to a higher interval like every 20 Minutes may help in some situations)



  3. Scheduler continues to "reschedule" Send Requests for a period of 7 days at as often an interval as the sender calculates for it to be.


  4. Sender calculates this next Schedule time by using the following formula:

    
    current time + the connection heartbeat timeout( 15 min default for RAS) +
    the retry delay(1 min default)* the number of retries( 5 default) * 2.  = 
    current time + 55 min<BR/> 


  5. Scheduler processes this send request over again from the beginning, scanning all outboxes for a sender that has the earliest chance to get the package sent. If you are using just one address to each site services by only one sender, the same sender gets the request resubmitted to it for sending to begin sometime after 55 minutes from the current time.


Additional query words: prodsms smssvr


Keywords          : kbSMS120 kbSender 
Version           : winnt:1.2
Platform          : winnt 
Issue type        : kbinfo 

Last Reviewed: March 18, 1999