ios
ios copied to clipboard
Reliable distributed agreement service for the cloud
The output of [deploy3serversfailed](../tree/master/scripts/docker/deploy3serversfailed) (previously PR #21) is not as expected. After node0 fails, the client seems unable to recover.
Currently, if a request for an application state machine is malformed then it must be replicated by consensus system before being applied to the state machine and detected. Server instead...
Currently Ios only builds Linux due to use of syscalls such as Fallocate, Fdatasync and directI/O
Bug found on OSX running the 3 server deployment from example3.conf with test clients. ``` fatal error: concurrent map writes goroutine 51 [running]: runtime.throw(0x28a3b0, 0x15) /usr/local/go/src/runtime/panic.go:547 +0x90 fp=0xc820b175e0 sp=0xc820b175c8 runtime.mapassign1(0x1c60c0,...
Currently, setting up an Ios cluster locally can require many tabs/window and managing quite a few processes. This could be greatly improved using Docker or a tools like [Foreman](http://blog.daviddollar.org/2011/05/06/introducing-foreman.html).
Currently, we use JSON for serialising data as it was easy to implement and is human readable. This should be switched to [Protobuf](https://github.com/google/protobuf) or similar before a production deployment.
There are various memory issues with JSON marshal/unmarshal. At the moment, we do this either in a strange verbose manner or has been replicated by a placeholder to prevent errors...
Currently, NO-OPS are represented explicitly like all other log entries, this is not necessary and could be done much more efficiently
Currently, multiple deployments of Ios running across the same hosts risk interfering with each other. This could be address by assigning each deployment at random ID and checking this ID...
Add -fresh option to server command line to instruct the Ios server to ignore and overwrite persistent storage if present. This will save the need to run ```rm persistent_*``` before...