Found issue with AWS migration import via S3
Description
Environment
AFFECTED CS IDs
Activity
Peter Sylvester February 21, 2024 at 1:25 PM
That was my original thought, but after checking around online it seems like the only option was to submit a forum post as I don’t think I have the option to submit a ticket on my own as I’m not a partner / client. This is why I wanted to contact our team to see if there was any avenue we had to open a ticket.
Do we have such a method?
Yura Sorokin February 14, 2024 at 2:36 PM
Sorry, won’t tell you more than just mentioned - contact AWS support. This is definitely a problem on their side. I can only suggest to check other (newer) versions of Aurora and identify in which version this rds_replication_status table has this channel_name column. If you submit exactly what is wrong and where it was fixed / broken to AWS support ticket you will have much better chances of them taking this seriously.
Peter Sylvester February 14, 2024 at 2:09 PM
I was speaking with and he mentioned you might be able to help here? This is really an issue on the AWS side, but I’m not sure what our process is for reaching out to them
Peter Sylvester February 14, 2024 at 1:42 PM
Have we heard anything back from AWS on this? I can always test again to confirm if the issue is resolved once they are ready.
Thank you!
Aaditya Dubey January 14, 2024 at 5:56 PM
Hi
Thank you for the report.
Details
Assignee
UnassignedUnassignedReporter
Peter SylvesterPeter SylvesterNeeds QA
YesComponents
Affects versions
Priority
Medium
Details
Details
Assignee
Reporter
Peter SylvesterNeeds QA
Components
Affects versions
Priority
MediumSmart Checklist
Open Smart Checklist
Smart Checklist
Open Smart Checklist
Smart Checklist
Please find the scenario below:
The problem is that I have a client that is planning to migrate their Percona 5.7 data set into Aurora. This would include migrating data and then setting up external replication from their existing instance until they are ready for the failover. Given their very large data set size I'm testing a migration using a physical data import via S3. I'm testing the high level steps for this so I can propose this as a solution.
In my lab I have an AWS EC2 instance running Ubuntu Focal 20.04, Percona Server 5.7, and Xtrabackup 2.4. I created a few test tables and backed up my data set using xtrabackup. After preparing the backup I uploaded the data set to S3 and then used the AWS management web consul to create a new Aurora 2.11.3 (5.7 compatible). The import succeeded with no issues and I could log into the instance and see the data.
I then attempted to set up external replication and this is where I saw the problem. After using the set external master command I ran the command to start replication and got the following error.
I did a bit of digging and it looks like the associated stored procedure references the chanel_name column in the mysql.rds_replication_status table...
But the column does not exist in the table.
I attempted to alter this table to add the column, but as expected this was not allowed in Aurora.
As such I was not able to configure replication from an external source.
So this isn't a bug specifically with MySQL or Aurora, but whatever process sets up the
needed tables for external replication when an Aurora instance is created based on an
S3 import. It's possible this might also impact other types of Aurora instances, but the only
thing I have tested is launching one via an S3 import.
I also tested this with 2.11.4 and had the same problem.