On Mar 15, 2013, at 1:40 PM, Stephen Bartell <[email protected]> wrote:
> > On Mar 14, 2013, at 3:36 PM, Robert Newson <[email protected]> wrote: > >> Runaway processes are the very devil but the problem is not specific >> to CouchDB, there is no CouchDB mechanism for this just as there's no >> bash/python/ruby/perl method to limit a while(true){} loop. > > Totally makes sense. The folks at Silverline pitch an idea about "application containers" to try to manage this sort of situation. I've never used myself but the tech always sounded neat: https://silverline.librato.com/promo/application_management >> >> Highly conflicted documents are painful to update and read. I can't do >> anything about that today. > > Thanks for your feedback! We've talked about this a bit internally at Cloudant. Perhaps a more appropriate discussion for dev@, but I think there are possible enhancements one can make to CouchDB's handling of deleted edit branches that allow the server to prune them automatically once it knows that all of its prior replication peers have received the tombstone at the end of the branch. In a multi-master scenario you do run the risk of re-vivifying part of the branch when the other side pushes edits back to you, but I think it's a risk that most folks who have been subjected to the pain of heavily-conflicted documents would be willing to take. Adam
