> On 2010-10-20 16:49:52, Todd Lipcon wrote: > > seems OK, but we're adding couple extra ms of latency here on all of these > > calls. Is that going to be expensive for assigning lots of regions? > > It seems we should be optimistic, and only really need to sync if we see > > unexpected state or the checked put fails? > > Jonathan Gray wrote: > Yeah, gave that a quick shot. It's not easy (the code gets messy quick > so it needs to be well thought out). > > I'd like to commit this and we can open another jira to deal with the > optimistic approach. > > Todd Lipcon wrote: > OK, that sounds good by me. We'll fix the perf issue in a follow-up. > > Would still like Stack to review, I don't know the master code well > enough to know if any other places might be missing, but I agree in concept :)
Cool. I'm kicking off a test suite run with this patch and my hbck patch right now as well. HBASE-3137 filed for perf improvements. - Jonathan ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: http://review.cloudera.org/r/1056/#review1590 ----------------------------------------------------------- On 2010-10-20 16:47:05, Jonathan Gray wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > http://review.cloudera.org/r/1056/ > ----------------------------------------------------------- > > (Updated 2010-10-20 16:47:05) > > > Review request for hbase, Todd Lipcon and stack. > > > Summary > ------- > > Adds a sync(path) operation into ZKW and three calls into it from the CAS > operations in ZKAssign. > > > This addresses bug HBASE-3136. > http://issues.apache.org/jira/browse/HBASE-3136 > > > Diffs > ----- > > trunk/src/main/java/org/apache/hadoop/hbase/zookeeper/ZKAssign.java 1025790 > trunk/src/main/java/org/apache/hadoop/hbase/zookeeper/ZooKeeperWatcher.java > 1025790 > > Diff: http://review.cloudera.org/r/1056/diff > > > Testing > ------- > > Still need to test more. I'm not sure it's possible (or feasible in a > reasonable amount of time) to make a unit test for this. We'd probably need > to dig into ZK or mock the hell out of stuff. > > > Thanks, > > Jonathan > >
