Conference forty2::x500

973.0. "Access problems from MB400" by CHEFS::BROWN_D () Thu Feb 13 1997 14:39

I have a UNIX server who's sole activity is running MAILbus400 and X.500 DSA 
    with supporting DECnet/OSI and X.25 comms.  The MTA receives a steady flow 
    of messages (10 per minute) each addressed to the same single o/r address 
    (a distribution list).  This distribution list is split by the MTA/DSA to 
    7 sub distribution lists and each one of these then produces approximately 
    8 recipients.  The MTA finally transfers out 16 messages each carrying 
    between 2 and 10 recipient addresses, each message going to a different 
    end MB400 system.

The DSA holds two naming contexts, is not the master of either but is a 
    supplier to 15 end sites.  The DIT is quite small, 600KB, and small 
    updates typically happen two to three times per day.

Communication between MTA and DSA is RFC1006.  Versions are MTA V1.4.5, 
    DSA V2.0A, UNIX V3.2D.

I get three problems on this system:

1. Every 10 minutes or so the MTA cannot gain access to the DSA and multiple 
    "Directory Service Error - Communication problem" events are generated by 
    the MTA. Often the DSA will also generate an "Authentication Failure - 
    Unknown User" event.  When the MTA tries again it must succeed as all 
    messages are delivered.

2. Rapid system memory wiring occurs.  The system needs rebooting every couple 
    of weeks to re-gain space.

3. At times the DSA process totally absorbs cpu time.  This occurs only on 
    certain days and then happens a few times in that day and lasts for about 
    5 minutes per time.

I have been looking at the DSA access rate.  This seems surprisingly high (to 
    me anyway for the number of messages) though I believe it is still within 
    performance capabilities.

Figures are as follows:

               /hr       /min
Reads          220,000   3666
Attrib.Errors   17,000    283
Name Errors         82    1to2
Results        203,000   3383

The customer is complaining about each of these problems and we need to 
    resolve to gain final acceptance.  It feels to me that something is 
    reaching limits in the MTA-DSA interaction causing the failed look-ups and 
    possibly leaving failed links around contributing to the memory wiring?

I would be grateful for any views on the above.
Dennis Brown

Dennis,

	I think this should be handled as a support call.

	Julian