Chuntao Hong

Peer xxx failed to catch up

3

follower重启后有一定概率出现这个错误，然后就加不进去 ```bash node lgraph:192.168.1.111:19091:0 can't do pre_vote as it is not in 192.168.1.111:19092:0,192.168.1.111:19093:0 Failed to join group: [E10003][C317683200A33A3ECD0844F9EC364D80][E10003]Peer 192.168.1.111:19091:0 failed to catch up ``` leader端消息： ```bash Receive AddPeerRequest to lgraph:192.168.1.111:19092:0...

automatically detect GPU compute capability and set -arch correctly

2

Currently we set -arch sm_35, which leads to "invalid device function" in older GPUs. There seems to be some effort in automatically detecting GPU compute capability and set it correctly:...

merging into mxnet repo

6

I propose to merge mxnet.cpp into mxnet project. 1. It will attract more attention to mxnet.cpp. 2. It will enable integrated test with mxnet, so that future checkins to mxnet...

[Feature Request] java binding of plugin api

1

**Is your feature request related to a problem? Please describe.** Currently only Python and C++ are supported as plugin programming language. **Describe the solution you'd like** Want a Java binding...

creates libminervaps.a which can be used by minerva to integrate with PS

Windows build

3

Now that mxnet depends on ps-lite, we should also have a Windows build for ps-lite, or mxnet won't build.

[dev branch] Cmake error

1

Both Windows and Linux builds fail with cmake. (Of course you can use make on Linux, I just tried cmake on Linux to verify the problem.) > CMake Error at...

Add package check

1

1. Add a magic number to the package header so when some other clients mistakenly connects to our server, we can reject the connection. 2. Add checksum to the package...

Commit 139b90ddaa7f79aa53e4f31649b166aea165a6f8 had a working MPI communication layer. But we removed it. Should be able to bring it back with moderate effort.