0

I have a fairly small database (2.5 MB dump) to migrate from AWS EC2 instance to Aurora (mariadb 10 to Aurora 5.7) Migration sticks on one of the tables. Even though the table has only ~1300 rows, total number of rows in the tables statistics keeps growing (more than 800 million!) And migration never finishes. But in fact the database is migrated. Just Full Load Rows Keeps increasing and migration never ends.

The problem is blocking us from migrating all the databases at once. In fact with setting 6 concurrent table migrations some of the tables stick and a large replication instance with 30 GB of memory, runs out of it!

I'm doing a Full load migration. In cloudwatch logs I see thousands of:

2019-04-09T18:52:06 https://forums.aws.amazon.com/I: Table 'rls_acronym_test'.'hours_overrides' has 1321 rows. Resume fetch is enabled (mysql_endpoint_unload.c:187)

John Rotenstein
  • 241,921
  • 22
  • 380
  • 470
hpaknia
  • 2,769
  • 4
  • 34
  • 63
  • Wondering if you ever found a solution for this. I'm experiencing the same thing - "Full Load Rows" keeps increasing, and never finishes.The logs are saying "Resume fetched succeeded, fetching more rows". Then followed by "Fetch received no date. Trying to resume fetch". – Michael LoCicero Feb 25 '20 at 19:24
  • The full LOB mod does not work. Use the `Limited LOB mode` with a very high value, a value you know your data for a row (or a cell maybe) does not exceed in size. If it worked as expected please answer so I can mark as the correct answer. - IMHO if you are migrating and replicating mariadb or mysql I strongly suggest to use native replication. AWS DMS is an engineering nightmare. – hpaknia Feb 26 '20 at 04:22

0 Answers0