Not as far as I know (maybe somewhere in the depths of Jira though...). Its about the fact that CP cross-table writes (e.g. mutable index writes) can be blocked by consumed handlers on another server. If that server also hosts a table to be indexed, it could want to write back to the first server, which is blocking an RPC handler for the original cross table write. We just need a way to specify custom RPC handlers for indexing (I think most of this work is already done, but its been a while since I've looked at that code) and then to actually use that handler from an HTable (special config option maybe?).
------------------- Jesse Yates @jesse_yates jyates.github.com On Thu, Apr 17, 2014 at 11:27 AM, Andrew Purtell <apurt...@apache.org>wrote: > On Thu, Apr 17, 2014 at 10:26 AM, James Taylor <jamestay...@apache.org > >wrote: > > > If possible, it'd be good to target this one too: > > - prevent deadlock during index maintenance ( > > https://issues.apache.org/jira/browse/PHOENIX-938). This one would be a > > 4.x > > only fix and will likely require HBase changes, so we'll likely need to > > wait until our next release if the changes don't make it into an HBase > > release prior to us releasing. > > > > Tell us more James. Is there an HBase JIRA filed for this? > > > > -- > Best regards, > > - Andy > > Problems worthy of attack prove their worth by hitting back. - Piet Hein > (via Tom White) >
