同一条记录多次修改分支回滚时执行顺序异常导致update time不一致出现脏写回滚失败

[vincent50411]

**环境说明：**生产环境中业务服务和seata server都是单节点部署，业务表中存在update time时间戳(6)，使用postgresql 15数据库。
**业务场景：**业务服务A更新A表一条记录得字段A_1后，调用业务服务B更新B表数据，业务服务B又调用A服务更新A表同一条记录的字段A_2，当业务执行异常后，执行全局回滚，业务服务A中提示全局回滚失败，日志记录显示：
2025-12-10 11:13:57.725 INFO 1 TID:N/A --- [rpcDispatch_RMROLE_1_2_16] o.a.s.c.r.p.c.RmBranchRollbackProcessor : rm handle branch rollback process:BranchRollbackRequest{xid='10.232.68.199:8091:5567173722488930773', branchId=5567173722488930775, branchType=AT, resourceId='jdbc:postgresql://xxxx:54321/haic', applicationData='null'}
2025-12-10 11:13:57.725 INFO 1 TID:N/A --- [rpcDispatch_RMROLE_1_2_16] o.a.seata.rm.AbstractRMHandler : Branch Rollbacking: 10.232.68.199:8091:5567173722488930773 5567173722488930775 jdbc:postgresql://xxxxx:54321/haic

2025-12-10 11:13:57.729 INFO 1 TID:N/A --- [rpcDispatch_RMROLE_1_2_16] o.a.s.r.d.u.AbstractUndoExecutor : Field not equals, name update_datetime, old value 2025-12-10 11:13:57.671, new value 2025-12-10 11:13:57.7

2025-12-10 11:13:57.730 ERROR 1 TID:N/A --- [rpcDispatch_RMROLE_1_2_16] o.a.s.r.d.DataSourceManager : branchRollback failed. branchType:[AT], xid:[10.232.68.199:8091:5567173722488930773], branchId:[5567173722488930775], resourceId:[jdbc:postgresql://xxxxx:54321/haic], applicationData:[null]. reason:[Branch session rollback failed because of dirty undo log, please delete the relevant undolog after manually calibrating the data. xid = 10.232.68.199:8091:5567173722488930773 branchId = 5567173722488930775]
2025-12-10 11:13:57.730 INFO 1 TID:N/A --- [rpcDispatch_RMROLE_1_2_16] o.a.seata.rm.AbstractRMHandler : Branch Rollbacked result: PhaseTwo_RollbackFailed_Unretryable

问题说明：业务服务A更新同一条记录2次，对应注册2个branch id，业务库中undo log表中存在2条log，branch id不同。日志显示update_datetime, old value 2025-12-10 11:13:57.671, new value 2025-12-10 11:13:57.7不一致，与undo log表对比，可以确认是rm回滚时，先回滚了最早修改的第一条记录，导致update time与当前记录不一致； 但是seata server未启用并行发送消息，branchsession 也做了逆序循环，理论上rm端应该是先收到最后一条branch才对，但是现象是发生了乱序。请问还有什么问题可能导致tc发送给rm的2条branch乱序？
seata server端的配置：
store.db.datasource=druid
store.db.dbType=postgresql
store.db.url=jdbc:postgresql://xxxxx:54321/seata
store.db.user=xxxx
store.db.password=xxxxx
store.db.driverClassName=org.postgresql.Driver
store.db.lockTable=lock_table
store.db.globalTable=global_table
store.db.queryLimit=100
store.db.maxConn=400
client.rm.lock.retryInterval=100
store.db.maxWait=5000
store.db.minConn=10
store.mode=db
store.db.branchTable=branch_table
service.vgroupMapping.my_tx_group=default

seata 客户端配置：
seata:
enabled: true
enable-auto-data-source-proxy: true
tx-service-group: my_tx_group
registry:
type: nacos
nacos:
application: seata-server
server-addr: ${common.nacos.server-addr}
namespace: ${common.nacos.discovery-namespace}
group: ${common.nacos.discovery-group}
cluster: default
username: ${common.nacos.discovery-username}
password: ${common.nacos.discovery-password}
config:
type: nacos
nacos:
server-addr: ${common.nacos.server-addr}
namespace: ${common.nacos.config-namespace}
group: ${common.nacos.discovery-group}
cluster: default
username: ${common.nacos.discovery-username}
password: ${common.nacos.discovery-password}
dataId: haic-seata.properties

[maple525866]

Is the time precision set by branch_table accurate enough?

[vincent50411]

branch_table表是官方脚本创建，实际存储值精度到3位毫秒值

[funky-eyes]

branch_table表是官方脚本创建，实际存储值精度到3位毫秒值

You should file an issue rather than a discussion, and include the version of the Seata server you are running.

[vincent50411]

seata server是2.2.0，客户端sdk 2.2.0，因为不太确认是不是问题，所以提了一个公开讨论话题

[funky-eyes]

Please provide the logs showing the rollback sequence on the server side. There should be no rollback sequence issues in version 2.2.0.

[vincent50411]

问题已经找到，生产环境版本用的是seata server 2.0.0版本，DefaultCore中的doGlobalRollback方法内查询branch session队列未做逆序处理：List branchSessions = globalSession.getSortedBranches(); 导致同一条记录多次修改后的分支回滚顺序是按照branch table表中的gmt create做asc排序的，即最早注册的分支事务最早回滚，出现了和当前最新记录比对时的字段不一致问题。2.2.0版本中已经处理了逆序。感谢

[vincent50411]

但是用2.2.0版本启动，seata server一直提示包太大解码失败的问题：是什么配置不对吗
2025-12-16 17:48:06.546 ERROR --- [NettyServerNIOWorker_1_4_4] [org.apache.seata.core.rpc.netty.MultiProtocolDecoder] [decode] []: Decode frame error, cause: Adjusted frame length exceeds 8388608: 539978086 - discarded
2025-12-16 17:48:06.547 WARN --- [NettyServerNIOWorker_1_4_4] [io.netty.channel.DefaultChannelPipeline] [onUnhandledInboundException] []: An exceptionCaught() event was fired, and it reached at the tail of the pipeline. It usually means the last handler in the pipeline did not handle the exception.
==>
io.netty.handler.codec.DecoderException: org.apache.seata.core.exception.DecodeException: io.netty.handler.codec.TooLongFrameException: Adjusted frame length exceeds 8388608: 539978086 - discarded
at io.netty.handler.codec.ByteToMessageDecoder.callDecode(ByteToMessageDecoder.java:499) ~[netty-codec-4.1.101.Final.jar:4.1.101.Final]
at io.netty.handler.codec.ByteToMessageDecoder.channelRead(ByteToMessageDecoder.java:290) ~[netty-codec-4.1.101.Final.jar:4.1.101.Final]
at io.netty.channel.AbstractChannelHandlerContext.invokeChannelRead(AbstractChannelHandlerContext.java:444) [netty-transport-4.1.101.Final.jar:4.1.101.Final]
at io.netty.channel.AbstractChannelHandlerContext.invokeChannelRead(AbstractChannelHandlerContext.java:420) [netty-transport-4.1.101.Final.jar:4.1.101.Final]
at io.netty.channel.AbstractChannelHandlerContext.fireChannelRead(AbstractChannelHandlerContext.java:412) [netty-transport-4.1.101.Final.jar:4.1.101.Final]
at io.netty.handler.timeout.IdleStateHandler.channelRead(IdleStateHandler.java:286) [netty-handler-4.1.101.Final.jar:4.1.101.Final]
at io.netty.channel.AbstractChannelHandlerContext.invokeChannelRead(AbstractChannelHandlerContext.java:442) [netty-transport-4.1.101.Final.jar:4.1.101.Final]
at io.netty.channel.AbstractChannelHandlerContext.invokeChannelRead(AbstractChannelHandlerContext.java:420) [netty-transport-4.1.101.Final.jar:4.1.101.Final]
at io.netty.channel.AbstractChannelHandlerContext.fireChannelRead(AbstractChannelHandlerContext.java:412) [netty-transport-4.1.101.Final.jar:4.1.101.Final]
at io.netty.channel.DefaultChannelPipeline$HeadContext.channelRead(DefaultChannelPipeline.java:1410) [netty-transport-4.1.101.Final.jar:4.1.101.Final]
at io.netty.channel.AbstractChannelHandlerContext.invokeChannelRead(AbstractChannelHandlerContext.java:440) [netty-transport-4.1.101.Final.jar:4.1.101.Final]
at io.netty.channel.AbstractChannelHandlerContext.invokeChannelRead(AbstractChannelHandlerContext.java:420) [netty-transport-4.1.101.Final.jar:4.1.101.Final]
at io.netty.channel.DefaultChannelPipeline.fireChannelRead(DefaultChannelPipeline.java:919) [netty-transport-4.1.101.Final.jar:4.1.101.Final]
at io.netty.channel.nio.AbstractNioByteChannel$NioByteUnsafe.read(AbstractNioByteChannel.java:166) [netty-transport-4.1.101.Final.jar:4.1.101.Final]
at io.netty.channel.nio.NioEventLoop.processSelectedKey(NioEventLoop.java:788) [netty-transport-4.1.101.Final.jar:4.1.101.Final]
at io.netty.channel.nio.NioEventLoop.processSelectedKeysOptimized(NioEventLoop.java:724) [netty-transport-4.1.101.Final.jar:4.1.101.Final]
at io.netty.channel.nio.NioEventLoop.processSelectedKeys(NioEventLoop.java:650) [netty-transport-4.1.101.Final.jar:4.1.101.Final]
at io.netty.channel.nio.NioEventLoop.run(NioEventLoop.java:562) [netty-transport-4.1.101.Final.jar:4.1.101.Final]
at io.netty.util.concurrent.SingleThreadEventExecutor$4.run(SingleThreadEventExecutor.java:997) [netty-common-4.1.101.Final.jar:4.1.101.Final]
at io.netty.util.internal.ThreadExecutorMap$2.run(ThreadExecutorMap.java:74) [netty-common-4.1.101.Final.jar:4.1.101.Final]
at io.netty.util.concurrent.FastThreadLocalRunnable.run(FastThreadLocalRunnable.java:30) [netty-common-4.1.101.Final.jar:4.1.101.Final]
at java.lang.Thread.run(Thread.java:750) [na:1.8.0_342]
Caused by: org.apache.seata.core.exception.DecodeException: io.netty.handler.codec.TooLongFrameException: Adjusted frame length exceeds 8388608: 539978086 - discarded
at org.apache.seata.core.rpc.netty.MultiProtocolDecoder.decode(MultiProtocolDecoder.java:132) ~[seata-core-2.2.0.jar:2.2.0]
at io.netty.handler.codec.LengthFieldBasedFrameDecoder.decode(LengthFieldBasedFrameDecoder.java:333) ~[netty-codec-4.1.101.Final.jar:4.1.101.Final]
at io.netty.handler.codec.ByteToMessageDecoder.decodeRemovalReentryProtection(ByteToMessageDecoder.java:529) ~[netty-codec-4.1.101.Final.jar:4.1.101.Final]
at io.netty.handler.codec.ByteToMessageDecoder.callDecode(ByteToMessageDecoder.java:468) ~[netty-codec-4.1.101.Final.jar:4.1.101.Final]
... 21 common frames omitted
Caused by: io.netty.handler.codec.TooLongFrameException: Adjusted frame length exceeds 8388608: 539978086 - discarded
at io.netty.handler.codec.LengthFieldBasedFrameDecoder.fail(LengthFieldBasedFrameDecoder.java:507) ~[netty-codec-4.1.101.Final.jar:4.1.101.Final]
at io.netty.handler.codec.LengthFieldBasedFrameDecoder.failIfNecessary(LengthFieldBasedFrameDecoder.java:493) ~[netty-codec-4.1.101.Final.jar:4.1.101.Final]
at io.netty.handler.codec.LengthFieldBasedFrameDecoder.exceededFrameLength(LengthFieldBasedFrameDecoder.java:377) ~[netty-codec-4.1.101.Final.jar:4.1.101.Final]
at io.netty.handler.codec.LengthFieldBasedFrameDecoder.decode(LengthFieldBasedFrameDecoder.java:423) ~[netty-codec-4.1.101.Final.jar:4.1.101.Final]
at org.apache.seata.core.rpc.netty.MultiProtocolDecoder.decode(MultiProtocolDecoder.java:105) ~[seata-core-2.2.0.jar:2.2.0]
... 24 common frames omitted