Pomelo.EntityFrameworkCore.MySql icon indicating copy to clipboard operation
Pomelo.EntityFrameworkCore.MySql copied to clipboard

Published 20 hours ago • verified publisher icon PomeloFoundation

Possible issues if NDBCluster engine is used

Open unsafePtr opened this issue 2 years ago • 0 comments

The issue

I am wondering what kind of issues we might have if we use NDBCluster engine? I understand that NDBCLUSTER is not within supported engines and I guess there are no plans for it. As an example regarding the limitations I see that only read commited isolation level is supported. How this can affect migrations?

Further technical details

We have tried to test migrations over NDBCluster. What we did:

  1. Always set maxLength for string columns (text type is not supported)
  2. Set on query execution engine=NDBCLUSTER at Db level.

We experience only one issue when running the migrations. On first run, everything runs properly without issues. However, when you try to re-run migration we see in the output Table x already exists. It looks like runner doesn't see Migrations table and rows within it.

I hope it will help to understand and mitigate issues if someone will decide to go with NDBCluster.

unsafePtr avatar Jan 27 '23 11:01 unsafePtr