[MDEV-13383] Spider is not maintaining SQL_MODE across nodes Created: 2017-07-26 Updated: 2019-04-30 Resolved: 2019-04-30 |
|
| Status: | Closed |
| Project: | MariaDB Server |
| Component/s: | Storage Engine - Spider |
| Affects Version/s: | 10.2.6 |
| Fix Version/s: | N/A |
| Type: | Bug | Priority: | Major |
| Reporter: | Anders Karlsson | Assignee: | Kentoku Shiba (Inactive) |
| Resolution: | Duplicate | Votes: | 1 |
| Labels: | None | ||
| Environment: |
Linux |
||
| Issue Links: |
|
||||||||
| Description |
|
When session level SQL_MODE is set in the spider node, this is not replicated to the data nodes in Spider, causing issues with data that is correct in the spider node SQL_MODE setting but is not in data nodes unless SQL_MODE is set there also to align with the spider node setting. |
| Comments |
| Comment by Simon Mudd [ 2017-07-26 ] |
|
This may not be noticed unless the spider node (or datanodes) have a stricter sql_mode setting than expected. e.g. during replication the upstream master could be configured differently. I saw this break replication. The issue was due to a configuration mistake but it's clear that sql_mode settings need to be applied on the data nodes to be the same as the session. Whether that comes through via a normal query or via replication won't matter. |
| Comment by David Thompson (Inactive) [ 2017-07-26 ] |
|
This should obviously be configurable since there is no strict requirement that the spider node is the same exact version or distribution as the data nodes so it's possible that sql_mode valid value support may differ. |
| Comment by Jacob Mathew (Inactive) [ 2017-08-24 ] |
|
Downgraded priority since the customer has a workaround. |