[GitHub] [activemq-artemis] gtully commented on pull request #3289: ARTEMIS-2918 Support Server restart on I/O errors

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

gtully commented on pull request #3289:
URL: https://github.com/apache/activemq-artemis/pull/3289#issuecomment-704410462


   I wonder if this feature could be recast as "keep the broker alive on JDBC io errors", a much narrower focus.
   
   Keeping a general broker alive in a jvm without a jvm restart will be very difficult, b/c errors can occur at any point during start and stop!
   
   If the focus is just JDBC, an essentially atomic storage, it could be possible to report JDBC errors to the clients and keep the broker alive.
   A client would see an exception and/or closed connection and retry in the normal way.
   The broker would resume normal service when it again was able to get a pooled JDBC connection.
   
   Note: the lease locker will need to deal with the potential of a jdbc pool to block, it will need to track the lease duration locally such that it can expire and stop if the lease is lost. In other words, the lease will only survive jdbc errors for the remainder of its lease.
   


----------------------------------------------------------------
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