[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference iosg::all-in-1

Title:ALL-IN-1 (tm) Support Conference
Notice:Please spell ALL-IN-1 correctly - all CAPITALS!
Moderator:IOSG::PYECE
Created:Fri Jul 01 1994
Last Modified:Fri Jun 06 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:2716
Total number of notes:12169

2654.0. "File Cabinet Server error -%RMS-F-EXENQLM" by WOTVAX::16.183.112.106::wotvax.reo.dec.com::fitzpatrick (Wisest is she who knows she does not know) Wed Apr 30 1997 21:12

 Hi,  Software:  ALL-IN-1 V3.2	    
		 VMS V6.2	    
		 TeamLinks V2.7 EC001, TeamLinks V2.1 EC005
                 & TeamLinks V2.1			  

 A customer is receiving the following error message in the file    
 cabinet server log file.    
 
	%RMS-F-EXENQLM exceeded enqueue quota 

 Currently their ENQLM (OAFC$SERVER_ENQLM) is set to 14,096. They have 3
 nodes in a cluster - 1 is a VAX, the other two are Alphas.   
 There are approximately 25 TeamLinks connections to the  VAX and about  
 340 per  Alpha. 

 The problem occurred on one of the Alphas, it had 250 TeamLinks 
 connections at the time.  The other Alpha had 334 connections at the 
 same time and was processing fine. Each customer has approximately
 10 drawers, 8 of which are shared  drawers. 

 The customer realises that in order to fix this problem they must 
 increase the ENQLM quota.  The problem is  - what should they increase 
 it to in order to avoid this problem happening again. Should any other 
 parameters be increased to help avoid this problem? 

 The File Cabinet Server is configured as follows: 

 Server Attributes: 

 Authorized Timeout: 600	Max Client Connects: 512 
 Distribution: OFF		Max # of Drawers: 250 
 Drawer Cache: 50		Port Number: 7373 
 Drawer TimeOut: 600		Session Timeout: 1200 
 Drawer Collisions: 0  

 Thanks in advance, 
 
 Emer
T.RTitleUserPersonal
Name
DateLines
2654.1IOSG::STANDAGEMy hovercraft is full of eelsThu May 01 1997 10:3920
    
    Emer,
    
    The V3.2 documentation should provide the necessary calculations for
    ENQLM, although as you've abviously increased the limit somewhat
    already perhaps you've already read that book !  I'll get our
    performance engineer to take a look at your query.
    
    Do you only run one FCS per node ?  Although the FCS can accept up to
    512 connections, it works at it's best between 200 and 300 clients. So,
    if one server is taking up to 380 connections, you might want to
    consider creating a new server and getting some of the clients to
    connect to that. This would also decrease the demand for ENQLM within
    the single process.
    
    -Kevin.
    
    
    
    
2654.2IOSG::DAVISThu May 01 1997 11:5520
    
       Your current ENQLM seems reasonable - but there may be some factors 
       at work that are using it up. For example has the number of drawers 
       on the system been underestimated.
    
       The ENQLM is the number of locks that the server can hold at one 
       time. This usually relates to the number of files that are open 
       which in turn is influenced by the number of drawers that are open or 
       in cache. 
    
       Anyway whatever the cause is of your high ENQ count there is no 
       penalty involved in raising it to a much higher value. Why not raise 
       it to 50,000.
    
       There are no other parameters that should be changed - but if the 
       ENQLM is being reached because the number of drawers have been 
       underestimated then you should also look at FILLM.
    
       					Mark