ctubbsii commented on issue #3371: URL: https://github.com/apache/accumulo/issues/3371#issuecomment-1546532083
> From an operational standpoint. I cannot have a scan thread that exposes a problem taking down tservers, especially when it is likely that that scan is being spread across a good portion of the system. Understood. However, from another operational standpoint, we can't have a bad scan thread corrupt a tservers' internal state and leave it running to cause more harm, like corrupting user data or returning bad results, either. I've seen that too. These are competing principles, but both have merits. The greater risk of the two really depends on the specific application and user risk tolerances. Best we can do is to try to make it possible for users to make their own choices, by making things more pluggable, which is what we've been doing over the last several years, to make Accumulo's components more modular, with SPI endpoints, and less tightly coupled default internal components. -- 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. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
