Jason Smith <jhs@...> writes: > > Thanks Paul! You have no idea how grateful I am for this. I was about to give up > > on CouchDB because there wasn't decent ubuntu packages. The idea of running it > > from a build-couchdb folder somewhere in production sounds horrible, > > Horrible? > > > and > > building from source is great until it's time to upgrade. > > For all its faults, build-couchdb **IS** building from source. Often > people upgrade by building a new release with a different `install=` > directory, so you have side-by-side installs, and you choose which > /path/to/couchdb to run. > > Having said that, I agree with your main point that building from > source is no longer a good expenditure of time and energy these days.
Hi Jason, I wanted to clarify on the meaning of "horrible", haha. I think build-couchdb is pretty useful. Yesterday I built it from source with build-couchdb and it worked great and was very simple. When I said it seems "horrible in production" it's because (I hope) I don't need multiple versions of CouchDB on my prod server. I want one version, and I need to know that when I upgrade to the next release all my config files, data files, log files, and init.d scripts will remain intact be used by the new version automatically. Unless I'm missing something, build-couchdb doesn't (easily) offer that, but an ubuntu package does a superb job of it. Anyway, I've now got some data in my CouchDB so it'll be interesting to see how it works. I'm excited about the master-master eventually consistent design. It's a necessity for those of us with multiple datacenters and web-servers in each datacenter that need to write to a local data-store.
