8.1.7.4 + 2 patches hangs on imp, strange waits

All:

I'm getting ready to move from 8.1.7.2 to 8.1.7.4. Since the vendor of the app I support is heavily dependent on exp/imp I'm sensitive to those issues. So on my test box (4 processor hp K-class, autoraid), I applied the 8.1.7.4 patch set, applied the two patches (2598387 and 2410612 to fix large buffer and repetitive data bugs), ran catalog, catproc and catexp, did an import of a smaller schema with no problem, then started the import about 1030 Thursday. The import runs within a script that drops the user, coalesces the ts, creates the user and imports, I run the script in the background, normally takes about 16-20 hours on this box IIRC. 1700 I check, I see it appears to be working on the first big table (3 million rows). 0900 I check script output file, it's still on that table? Check top, imp's been spinning 100% cpu for hours, dbw hardly has any time at all? Data files haven't been written to since 1250? (norachivelog mode), logs haven't advanced since 1219, no errors in alert log?

  1 select a.sid, a.event, a.state, a.wait_time, a.seconds_in_wait, b.username
  2 from v$session_wait a,v$session b
  3* where a.sid = b.sid
SQL> /        SID EVENT
---------- ----------------------------------------------------------------

STATE                WAIT_TIME SECONDS_IN_WAIT USERNAME


------------------- ---------- ---------------

------------------------------

         1 pmon timer
WAITING                      0             238

         2 rdbms ipc message
WAITING                      0             308

         3 rdbms ipc message
WAITING                      0            2158


       SID EVENT


---------- ----------------------------------------------------------------

STATE                WAIT_TIME SECONDS_IN_WAIT USERNAME


------------------- ---------- ---------------

------------------------------

         6 rdbms ipc message
WAITING                      0          242837

         4 rdbms ipc message
WAITING                      0               0

         5 smon timer
WAITING                      0           69700


       SID EVENT


---------- ----------------------------------------------------------------

STATE                WAIT_TIME SECONDS_IN_WAIT USERNAME


------------------- ---------- ---------------

------------------------------

        11 SQL*Net message to client
WAITED UNKNOWN TIME         -2               0 SYSTEM

         7 SQL*Net message from client
WAITING                      0           77696 SYSTEM

8 rows selected.

SQL> Oh duh, forgot to start listener. Should imp care? What would smon, pmon and those ipc messages be about? smon couldn't be doing it's lmt thing (only sort ts is lmt, and patch requires several bounces, didn't do anything before import).

While I wait for customer to straighten out Oracle Corpse screw up of support contract yet again, anyone have any ideas?

imp uses: fromuser touser ignore=y commit=y buffer=20000000 Yes, exp uses compress=y, I do want to go to lmt's which is one of the reasons I want to go to 8.1.7.4.

After bouncing the db, I still see those waits (with the seconds reset). What are they?

jg
--

@home.com is bogus.
http://garry.to Received on Fri Jul 11 2003 - 13:33:48 CDT