Skip to:

  1. Skip to Jira Navigation
  2. Skip to Side Navigation
  3. Skip to Main Content

Document mixed cluster write (PXC8 while PXC5.7 nodes are still part of the cluster) should not be completed.

Description

We should expand the documentation regarding upgrading from 5.7 to 8.0 .

https://www.percona.com/doc/percona-xtradb-cluster/8.0/howtos/upgrade_guide.html

If downtime is not possible, the rolling upgrade is supported but ensure the traffic is controlled during the upgrade and writes are directed only to 5.7 nodes until all nodes are upgraded to 8.0.

Replication from 8.0 to 5.7 is not supposed to work (and it doesn't). At the moment a user tries to write on 8.0 while still has 5.7 on the cluster it will break the cluster with:

This is especially a problem since ProxySQL by default uses an internal algorithm to chose its writer. Listing the errors on a limitations page or even on the upgrade documentation will safe support a lot of tickets .

Environment

None

Smart Checklist

Activity

Juan Arruti April 17, 2020 at 7:21 PM

Hi,

It can be mentioned super_read_only must be enabled when starting a 8.0 PXC node when there are still 5.7 nodes in the cluster. Another option is to set wsrep_reject_queries to ALL, however this is more restrictive since read-only queries won't be allowed as well.

Regards,

Done

Details

Assignee

Reporter

Needs Review

Yes

Time tracking

1h logged

Components

Fix versions

Affects versions

Priority

Smart Checklist

Created April 3, 2020 at 7:57 PM
Updated March 6, 2024 at 9:39 PM
Resolved April 29, 2020 at 11:51 AM