HCL
Skip to main content  
 
   


SPRTechnote


Repeated server crashes due to size of NameServer table

Technote Number: 1216813


Problem:
This issue was reported to Quality Engineering as SPR# ERYR6FEMZM and was fixed
in Notes/Domino releases 7.0.1, 6.5.5, and Domino 6.5.4 Fix Pack 3 (FP3).

Excerpt from the Lotus Notes and Domino MR releases 7.0.1, 6.5.5, and 6.5.4 FP3
fix lists (available at http://www.ibm.com/developerworks/lotus):

Networking & Dialup
SPR# ERYR6FEMZM - With this fix, the maximum number of servers supported in a
single domain has been doubled. Also an error message will be displayed if the
maximum number is exceeded.

If you haven't yet upgraded and experience this problem, contact Product
Support to see if a test fix is available for your particular configuration.

Background:
Server names and information about the servers are collected in a set of "in
memory" NameServer tables at server start time, one for each port. These tables
have no upper limit as to the number of servers supported.

Some server-side routines return a second different server name table
constructed by reading the port block tables. If the existing code exceeds 64K
in size, it truncates the second NameServer table and sometimes can cause
memory overwrites. Because this table holds server names and additional
information about servers, this table could be truncated with as few as 600-800
server names.

The fixed code will return a table with no truncation.

NOTE: There are other cases in which certain routines use the NameServer table
described above to construct a list of hierarchical server names. The server
returns a text list which is still limited to 64K, even though 64K can handle a
significant number of names (in one particular case, 2000).

In the event that the text list exceeds 64K, the server displays an error
message on startup. Additionally, the console command, "show nstable," can be
entered at any time to display the servers in the table and to check whether
the size of the list has been exceeded.

Cases which may evidence a problem:

1. When SmartUpgrade finds a down server, it searches the cluster for a
replica. If it can't get "clustermates" (a very rare event), it gets the full
list of servers.

2. Cluster analysis might report a non-existent problem if the server list is
truncated.

3. In certain cases, DECS (Domino Enterprise Connection Services) generates a
list of all servers which might get truncated.

4. A C++ API method used to return a list of all server name might return a
truncated list.

5. The ccMail migration process that populates a drop-down list with all
servers in the domain may return a truncated list.

6. The Notes NT user manager feature that populates a list box with all server
name might return a truncated list.

Refer to the Upgrade Central site for details on upgrading Notes/Domino.
More >





  Document options
Print this document
Print view

  Search
Search Advanced Search


  Fix list views

 RSS feeds   RSS
Subscribe to the fix list

  Resources
Using this database
View notices

  HCL Support
HCL Support


    About HCL Privacy Contact