EdColeman
EdColeman
I added a more general comment on #4409, While this may be useful in some cases, I'm not sure it solves the original issue. Mainly it seems that this would...
Would there be any benefit of allowing the property to be dynamic and stored in ZooKeeper? That would also require a watcher to react to updates - and that is...
I updated the initial comment. The goal is to terminate tservers that are running older versions - this is accomplished by removing their locks. It is possible to have the...
Tservers were specifically called out in https://github.com/apache/accumulo/issues/1300 - it may be possible / desirable to add other services if they are identified.
94b8f8b5d8 changes the logging to provide list of lock that were deleted as well as the locks that failed to be removed. The timeout was increased to virtually block -...
@ivakegg - wonder if you'd like to chime in on this. Do you think this is a good check to perform? Also, is it better to have it fail hard,...
With the current change - the check tries to contact the tservers and if any are found, then the list is printed at the end of the check, but the...
The pre-upgrade stage was added to support code / checks that are likely independent of a specific version. Checking the ZooKeeper authorizations is one example. It is not version dependent,...
NOTE to reviewers: I am not sure if this will address the problem - it will depend on how often the DefaultCompactionPlanner is instantiated because this fix will only work...
I have no issues with a more general solution. This can be closed if that's the way we want to go. My rational for this narrow PR was that, if...