Topic
No replies
SystemAdmin
SystemAdmin
17917 Posts
ACCEPTED ANSWER

Pinned topic SQL2036N when executing LOAD with HADR while active with no shared disk

‏2013-02-04T20:37:23Z |
Hi,

I have a HADR installation (two different machines, with different storage with the same structure), however there is not any shared device between the servers.

The users use LOAD to put data in the tables, but the remote database is NOT deactivated before this process. This means that the load instruction arrives in the standby, and the remote db cannot find the load file generating this error:


2013-02-03-22.08.13.845698-300 E1329297E363        LEVEL: Error PID     : 22232                TID  : 140472370587392PROC : db2sysc INSTANCE: db2inst1             NODE : 000 EDUID   : 1805                 EDUNAME: db2med.55.1 FUNCTION: DB2 UDB, database utilities, sqluMCCheckDevType, probe:98 MESSAGE : SQL2036N  The path 

for the file or device 
"" is not valid.


I now that the process should be (when no shared disk):
  • Deactivate remote db (stop hadr)
  • Perform the load.
  • Copy the generated file (SAMPLE.4.db2inst1.NODE0000.CATN0000.20130204130123.001) in the path
  • Start HADR.
(As explained here: http://www.db2dean.com/Previous/HADRLoad.html)

However, this process is not performed by the users, and some tablespaces in the remote database change their status to restore pending.

How can I put the database back online with all tablespace in normal without performing a full database restore?
How can I use the load generated file SAMPLE.4.db2inst1.NODE0000.CATN0000.20130204130123.001 in the remote database?
(Look at the 4 after the name, this means this file was generated by load.)