From @biljanaLukovic on Fri May 13 2016 13:31:28 GMT+0000 (UTC)
What is the recommended way to achieve high availability? I want to have two nodes exposing IPFS WEB service. If one node goes down I want to make sure that data is available through the other.
Thanks
Copied from original issue: https://github.com/ipfs/faq/issues/121
From @harlantwood on Sun May 15 2016 18:15:50 GMT+0000 (UTC)
Great question.
From @lgierth on Mon May 16 2016 00:01:12 GMT+0000 (UTC)
Regarding high availability of data, there are concrete plans for an ipfs-cluster tool: https://github.com/ipfs/notes/issues/58
High availability for the gateway can be achieved by e.g. round-robin DNS, or load balancing.
From @RichardLitt on Mon May 16 2016 13:50:33 GMT+0000 (UTC)
@lgierth What can @biljanaLukovic do right now to ensure high availability? Is there a guide?
From @hermanjunge on Tue May 17 2016 10:30:34 GMT+0000 (UTC)
+1 Currently I’m spinning IPFS nodes behind an Nginx server. I was recommended this package to achieve consensus among nodes. Haven’t checked it yet. Here it is:
Let me know what have you found so far, so we can work together on a first approach to solve this problem quickly.
From @salsa-dev on Sat May 21 2016 22:07:32 GMT+0000 (UTC)
+1 I’m also willing to build a service to provide data from libraries through ipfs gateway. The current problem is in synching all the data on two ipfs nodes and pushing data from other nodes. I’m aware of several solutions which haven’t tried yet:
- A consortium to help persist IPFS objects https://github.com/pipermerriam/ipfs-persistence-consortium
- PinCoop is a set of API endpoints and WebUI for adding public daemons, retrieve a list of all of them and to add hashes to be pinned on those public daemons. https://github.com/victorbjelkholm/pincoop
- IPFS replication service with one line of unix shell code! https://gist.github.com/kyledrake/e8e2a583741b3bb8237e