rmcp: fix transaction retry infinite loop

canteuni commented 10 months ago

This aims to fix issue #159 :

raise a RetryError instead of an "Unassigned variable" exception if we exceed the self.max_retries during the transaction loop
prevent from looping indefinitely when checking for mismatch between request header and received data if the ipmi message is not a bridged message

Perhaps it's better to use a TimeoutError for the transaction loop because we only increase the retry counter on socket.timeout exception, I'm not sure.

This can require further testing as it take care of a quite annoying to reproduce bug.

coveralls commented 10 months ago

coverage: 69.371% (-0.06%) from 69.434% when pulling 80823d53f1e6f627fd7f9b0c77d40a7ea7c68b9a on canteuni:fix_transaction_retry into 8e83f6f1de1c774cc9bdce90955b9917674351d4 on kontron:master.

hthiery commented 10 months ago

Unfortunatly I cannot test this change at the moment. So I will apply this PR without testing it now.

kontron / python-ipmi

rmcp: fix transaction retry infinite loop #160