update using case when causes fragmentation routing to drift

[tuyingDavid]

Bug Report

For English only, other languages will not accept.

Before report a bug, make sure you have:

• Searched open and closed GitHub issues.
• Read documentation: ShardingSphere Doc.

Please pay attention on issues you submitted, because we maybe need more details.
If no response anymore and we cannot reproduce it on current information, we will close it.

Please answer these questions before submitting your issue. Thanks!

Which version of ShardingSphere did you use?

4.1.1

Which project did you use? ShardingSphere-JDBC or ShardingSphere-Proxy?

ShardingSphere-JDBC

Expected behavior

routeValue = 'DS632624670761930754'

Actual behavior

routeValue = '2021-10-23 01:59:44.034'

Reason analyze (If you can)

Version 3.1.0 is no problem ，But upgrading to 4.1.1 fragmented routing has problems
I tried to experiment with several schemes，
Finally, tracking and debugging found，
If I only use the case when statement to update, there is no problem, if the non-case when statement is added, the fragmentation strategy will drift。

It feels like gmt_modify =? Caused by re-marking,
ParameterMarkerExpressionSegment(startIndex=339, stopIndex=339, parameterMarkerIndex=1)
Theoretically, parameterMarkerIndex should be 0

Steps to reproduce the behavior, such as: SQL to execute, sharding rule configuration, when exception occur etc.

#Sharding Rule config
shardingRule:
tables:
t_asset_detail:
actualDataNodes: ds$->{0..13}.t_asset_detail
databaseStrategy:
inline:
shardingColumn: asset_item_no
algorithmExpression: ds$->{Math.abs(asset_item_no.hashCode() % 14)}
bindingTables:
- t_asset_detail

Error example
UPDATE t_asset_detail
SET holder =
CASE WHEN
asset_item_no = ? AND ? >= gmt_holder THEN ? ELSE holder END ,
gmt_modify = ?
WHERE asset_item_no = ?
AND is_deleted = 0

Parameters: DS632624670761930754(String), 2021-10-23 01:50:21.416(Timestamp), 4(Integer), 2021-10-23 01:50:21.416(Timestamp), DS632624670761930754(String)
Normal example
UPDATE t_asset_detail
SET holder =
CASE WHEN
asset_item_no = ? AND ? >= gmt_holder THEN ? ELSE holder END
WHERE asset_item_no = ?
AND is_deleted = 0

Parameters: DS632624670761930754(String), 2021-10-23 01:50:21.416(Timestamp), 4(Integer), DS632624670761930754(String)

Example codes for reproduce this issue (such as a github link).

[lanchengx]

Hi @Totuying Thanks for your feedback, could you try it with the latest version

[tuyingDavid]

Hi @Totuying Thanks for your feedback, could you try it with the latest version

The 4.x version problem still exists

[zhaojinchao95]

Hi @Totuying Thanks for your feedback, could you try it with the latest version

The 4.x version problem still exists

you can try it with 5.x release

[tuyingDavid]

Hi @Totuying Thanks for your feedback, could you try it with the latest version

The 4.x version problem still exists

you can try it with 5.x release

tks ,Do not consider using the 5.x version for the time being ,We try to change the SQL order, Just move gmt_modify = ? to first row

Error example
UPDATE t_asset_detail
SET holder =
CASE WHEN
asset_item_no = ? AND ? >= gmt_holder THEN ? ELSE holder END ,
gmt_modify = ?
WHERE asset_item_no = ?
AND is_deleted = 0

The changed SQL is successful
UPDATE t_asset_detail
SET
gmt_modify = ?,`
holder =
CASE WHEN
asset_item_no = ? AND ? >= gmt_holder THEN ? ELSE holder END
WHERE asset_item_no = ?
AND is_deleted = 0

[tuyingDavid]

We found the real reason，In fact, shardingsphere does not support case when routing, so it happens that the first parameter happens to be the sharding key，This is not a bug。