Re: datapump export incredibly slow

  • From: David Fitzjarrell <oratune@xxxxxxxxx>
  • To: "Chris.Stephens@xxxxxxx" <Chris.Stephens@xxxxxxx>, "'oracle-l@xxxxxxxxxxxxx'" <oracle-l@xxxxxxxxxxxxx>
  • Date: Fri, 14 Feb 2014 14:10:45 -0800 (PST)

I have to ask why  you're running PARALLEL=4 and only creating one dump file?  
It may be that the parallel processes running for the export are waiting for a 
coordinator to consolidate query results and also the parallel write processes 
may get into a queue to write to the singleton dump file.  Such actions could 
very well result in the waits you're seeing where 'nothing' is being done.  

Can you try using as many dump files as parallel processes (4)?  Your parfile 
list would then look like this:


SCHEMAS=xxxxxxx
CONTENT=ALL
DUMPFILE=xxxxxxx_full1.dmp, xxxxxxx_full2.dmp, xxxxxxx_full3.dmp, 
xxxxxxx_full4.dmp 
LOGFILE=xxxxxxxx_full.exp.log
CLUSTER=N
PARALLEL=4
COMPRESSION=ALL
FLASHBACK_TIME=SYSTIMESTAMP
EXCLUDE=STATISTICS

This might mitigate the issues  you report.

 
David Fitzjarrell
Primary author, "Oracle Exadata Survival Guide"




On Friday, February 14, 2014 2:01 PM, "Stephens, Chris" 
<Chris.Stephens@xxxxxxx> wrote:
 
 
11.2.0.4 RAC on Oracle Linux 6.
 
Right now I’m battling an unbelievably slow export but imports seem 
unreasonably slow as well.  This occurs in two different RAC environments.  I 
have a ticket open with Oracle but that’s not getting me anywhere quite.
 
I’m exporting with the following parameters:
 
SCHEMAS=xxxxxxx
CONTENT=ALL
DUMPFILE=xxxxxxx_full.dmp
LOGFILE=xxxxxxxx_full.exp.log
CLUSTER=N
PARALLEL=4
COMPRESSION=ALL
FLASHBACK_TIME=SYSTIMESTAMP
EXCLUDE=STATISTICS
 
It looks like the export does some work (lots of direct path reads) for a while 
and then does nothing with tons of waits similar to:
 
*** 2014-02-14 14:51:25.924
WAIT #139990593434552: nam='wait for unread message on broadcast channel' ela= 
1000810 channel context=93982405280 channel handle=94519841120 p3=0 obj#=803670 
tim=1392411085924778
 
*** 2014-02-14 14:51:26.925
WAIT #139990593434552: nam='wait for unread message on broadcast channel' ela= 
1000867 channel context=93982405280 channel handle=94519841120 p3=0 obj#=803670 
tim=1392411086925789
EXEC 
#139990593434552:c=2999,e=5009559,p=0,cr=4,cu=0,mis=0,r=0,dep=3,og=4,plh=0,tim=1392411086926186
ERROR #139990593434552:err=25228 tim=1392411086926229
CLOSE #139990593434552:c=0,e=4,dep=3,type=3,tim=1392411086926370
 
I’m guessing that error possibly indicates a bug?
 
The export is estimated to be ~80GB.  I let it run for several hours this 
morning and it was nowhere close to being done.
 
Anybody have any ideas or seen this before?
CONFIDENTIALITY NOTICE:
This message is intended for the use of the individual or entity to which it is 
addressed and may contain information that is privileged, confidential and 
exempt from disclosure under applicable law. If the reader of this message is 
not the intended recipient
 or the employee or agent responsible for delivering this message to the 
intended recipient, you are hereby notified that any dissemination, 
distribution or copying of this communication is strictly prohibited. If you 
have received this communication in error,
 please notify us immediately by email reply.

Other related posts: