This was first published on https://blog.dbi-services.com/impdp-logtimeall-metricsy-and-12cr2-parallel-metadata (2017-11-28)
Republishing here for new followers. The content is related to the the versions available at the publication date
A quick post to show why you should always use LOGTIME=ALL METRICS=Y when using Data Pump. Just look at an example showing the timestamp in front of each line and a message about each task completed by the worker:
15-NOV-17 11:48:32.305: W-5 Processing object type DATABASE_EXPORT/SCHEMA/PROCACT_SCHEMA 15-NOV-17 11:48:34.439: W-13 Completed 28 PROCACT_SCHEMA objects in 1 seconds 15-NOV-17 11:48:34.439: W-13 Completed by worker 1 28 PROCACT_SCHEMA objects in 1 seconds 15-NOV-17 11:48:34.440: W-13 Processing object type DATABASE_EXPORT/SCHEMA/TABLE/TABLE 15-NOV-17 11:48:35.472: W-17 Startup took 70 seconds 15-NOV-17 11:48:35.596: W-18 Startup took 70 seconds 15-NOV-17 11:48:35.719: W-20 Startup took 70 seconds 15-NOV-17 11:48:35.841: W-19 Startup took 70 seconds 15-NOV-17 11:48:43.520: W-5 Completed 1714 TABLE objects in 7 seconds 15-NOV-17 11:48:43.520: W-5 Completed by worker 1 39 TABLE objects in 2 seconds 15-NOV-17 11:48:43.520: W-5 Completed by worker 2 113 TABLE objects in 7 seconds 15-NOV-17 11:48:43.520: W-5 Completed by worker 3 85 TABLE objects in 6 seconds 15-NOV-17 11:48:43.520: W-5 Completed by worker 4 111 TABLE objects in 6 seconds 15-NOV-17 11:48:43.520: W-5 Completed by worker 5 25 TABLE objects in 1 seconds 15-NOV-17 11:48:43.520: W-5 Completed by worker 6 113 TABLE objects in 7 seconds 15-NOV-17 11:48:43.520: W-5 Completed by worker 7 113 TABLE objects in 7 seconds 15-NOV-17 11:48:43.520: W-5 Completed by worker 8 111 TABLE objects in 6 seconds 15-NOV-17 11:48:43.520: W-5 Completed by worker 9 89 TABLE objects in 5 seconds 15-NOV-17 11:48:43.520: W-5 Completed by worker 10 74 TABLE objects in 4 seconds 15-NOV-17 11:48:43.520: W-5 Completed by worker 11 113 TABLE objects in 7 seconds 15-NOV-17 11:48:43.520: W-5 Completed by worker 12 113 TABLE objects in 7 seconds 15-NOV-17 11:48:43.520: W-5 Completed by worker 13 34 TABLE objects in 2 seconds 15-NOV-17 11:48:43.520: W-5 Completed by worker 14 111 TABLE objects in 6 seconds 15-NOV-17 11:48:43.520: W-5 Completed by worker 15 108 TABLE objects in 7 seconds 15-NOV-17 11:48:43.520: W-5 Completed by worker 16 90 TABLE objects in 4 seconds 15-NOV-17 11:48:43.520: W-5 Completed by worker 17 82 TABLE objects in 4 seconds 15-NOV-17 11:48:43.520: W-5 Completed by worker 18 40 TABLE objects in 3 seconds 15-NOV-17 11:48:43.520: W-5 Completed by worker 19 97 TABLE objects in 6 seconds 15-NOV-17 11:48:43.520: W-5 Completed by worker 20 53 TABLE objects in 3 seconds 15-NOV-17 11:48:44.266: W-1 Processing object type DATABASE_EXPORT/SCHEMA/TABLE/TABLE_DATA
Here, I was running an import with PARALLEL=20 and I can see exactly how many tables were processed by each worker. You see it is ‘TABLE’ and not ‘TABLE_DATA’ which is the proof that 12cR2 can import metadata in parallel.
I see no reason not to use LOGTIME=ALL METRICS=Y always and you will be happy to have this detail if something goes wrong.
Hi Franck, Thanks for the article, it’s one of the few around explaining the Parallel Metadata creation and METRICS=Y. I’ve found (and had accepted by Oracle Support) a serious bug in this area when large numbers of Sequences and/or Constraints exist. This bug means a schema with a large number of objects can no longer be exported/imported in parallel, it’s very frustrating. It’s too involved to write-up here, please drop me email if you’re interested and I’ll send the re-producible Test Case. Cheers