[ https://issues.apache.org/jira/browse/ACCUMULO-118?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13890330#comment-13890330 ]
Josh Elser commented on ACCUMULO-118: ------------------------------------- I think the short of it here: it's hard. I remember when Eric was initially working on absolute paths and I thought "hrm, that's a good idea. should simplify a lot of things in the end". In hindsight, I don't think I really considered all of the difficulties that the changes introduce (most notably around upgrades and namenode/namespace decommissioning). Maintaining a long-running feature branch isn't too bad as long as the code you tweaked also doesn't change out from underneath you. I agree with you Keith, I think that focusing on design docs before starting to work on it can help quite a bit on a couple of levels (avoid flaws in design, catch bugs earlier, net a better architected solutions). Additionally, firming up a design can also help us break down "really big" problems into "slightly less big" problems which will likely help manage those changes. I think we've generally tried to abstain from requiring voting, but if that's what we need to get eyes on ideas, so be it. If we can get good, thought-out reviews without voting (which I think we've been fairly decent at so far, but we haven't had "big" designs go through review yet), I'd rather stay that way. > accumulo could work across HDFS instances, which would help it to scale past > a single namenode > ---------------------------------------------------------------------------------------------- > > Key: ACCUMULO-118 > URL: https://issues.apache.org/jira/browse/ACCUMULO-118 > Project: Accumulo > Issue Type: Improvement > Components: master, tserver > Reporter: Eric Newton > Assignee: Eric Newton > Priority: Blocker > Fix For: 1.6.0 > > Attachments: ACCUMULO-118-01.txt, ACCUMULO-118-02.txt > > Original Estimate: 2,016h > Remaining Estimate: 2,016h > > Consider using full path names to files, which would allow the servers to > access the files on any HDFS file system. > Work may exist elsewhere to run HDFS using a number of NameNode instances to > break up the namespace. > We may need a pluggable strategy to determine namespace for new files. -- This message was sent by Atlassian JIRA (v6.1.5#6160)