[ https://issues.apache.org/jira/browse/CASSANDRA-7066?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14522990#comment-14522990 ]
Benedict commented on CASSANDRA-7066: ------------------------------------- If we're only keeping the tmp files for a bit of convenience for power users (no one else should be poking around in there), then I think the pain/problems it causes Windows, and the extra complexity it requires for CASSANDRA-6916 don't warrant their retention. [~JoshuaMcKenzie] [~krummas] care to weigh in? My view is users shouldn't generally be copying files around inside of a database unless they absolutely 100% know what they are doing. We provide them tools for producing and importing sstables, and they should be using those, generally, not mucking around on the file system directly. I may be missing some common operator behaviours that require it, though. Since the worst case outcome for operators is some duplicated data (or some sstables that are corrupt, and so they have to delete//scrub, if they somehow cause a partially written file to be retained), I don't see a major risk factor either. > Simplify (and unify) cleanup of compaction leftovers > ---------------------------------------------------- > > Key: CASSANDRA-7066 > URL: https://issues.apache.org/jira/browse/CASSANDRA-7066 > Project: Cassandra > Issue Type: Improvement > Components: Core > Reporter: Benedict > Assignee: Stefania > Priority: Minor > Labels: compaction > Fix For: 3.x > > > Currently we manage a list of in-progress compactions in a system table, > which we use to cleanup incomplete compactions when we're done. The problem > with this is that 1) it's a bit clunky (and leaves us in positions where we > can unnecessarily cleanup completed files, or conversely not cleanup files > that have been superceded); and 2) it's only used for a regular compaction - > no other compaction types are guarded in the same way, so can result in > duplication if we fail before deleting the replacements. > I'd like to see each sstable store in its metadata its direct ancestors, and > on startup we simply delete any sstables that occur in the union of all > ancestor sets. This way as soon as we finish writing we're capable of > cleaning up any leftovers, so we never get duplication. It's also much easier > to reason about. -- This message was sent by Atlassian JIRA (v6.3.4#6332)