[GitHub] [incubator-doris] WingsGo opened a new issue #3292: [Transaction] FE will not cancel some stream load txn when the stream load cannot be finished

[GitBox <gi...@apache.org>]

WingsGo opened a new issue #3292: [Transaction] FE will not cancel some stream load txn when the stream load cannot be finished
URL: https://github.com/apache/incubator-doris/issues/3292
 
 
   In current StreamLoad design, FE will recive load information from BE by RPC call, but I found that if the coordinator BE down when doing stream load fragment, FE will not receive any rollback txn call and the txns on the coordinator BE will never finished and the state  is always PREPARE,  it is because FE has no idea that the txn will aborted finally cause the fragment execute failed, and be cann't notify FE that case. So, if user set a long timeout stream load in parameters, there are a lots txns in GlobalTxnMgr, and they occupied the txn queue sizes and memory usage in FE, it may block the stream load.So, I define a callback function to cancel all txn which coordiator BE is down to solve this problem

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


With regards,
Apache Git Services

---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@doris.apache.org
For additional commands, e-mail: commits-help@doris.apache.org