Here's the list of commands I've used to run various processes (logcabin, coordinator, master) on various nodes on RAMCloud. Keeping them around for ease of use (by me and others, including Stephen.)
LogCabin:
I typically run the logcabin cluster on rc10, rc09, rc08.
I have scripts do-clean-initialize and do-reconfigure in ~/ramcloud/logcabin to make life a little easier. You can copy these to your logcabin directory. Alternately, you can follow the instructions in Additional notes -> LogCabin.
Before the first time you run this, you have to do the setup (step 0) described below in Additional notes -> LogCabin.
on some node x (doesn't matter which, so lets say rcmaster):
cd logcabin
./do-clean-initialze
on all nodes on which i want to run the logcabin cluster:
cd logcabin
./build/LogCabin
on x:
./do-reconfigure
Coordinator: on rc10:
./obj.master/coordinator -C "infrc:host=192.168.1.110,port=11114" -z "logcabin06:61023" -l "DEBUG"
Masters: on rc01 through rc06 (each command corresponds to each node):
./obj.master/server -L "infrc:host=192.168.1.101,port=11113" -C "infrc:host=192.168.1.110,port=11114" -M r=0
./obj.master/server -L "infrc:host=192.168.1.102,port=11113" -C "infrc:host=192.168.1.110,port=11114" -M r=0
./obj.master/server -L "infrc:host=192.168.1.103,port=11113" -C "infrc:host=192.168.1.110,port=11114" -M r=0
./obj.master/server -L "infrc:host=192.168.1.104,port=11113" -C "infrc:host=192.168.1.110,port=11114" -M r=0
./obj.master/server -L "infrc:host=192.168.1.105,port=11113" -C "infrc:host=192.168.1.110,port=11114" -M r=0
./obj.master/server -L "infrc:host=192.168.1.106,port=11113" -C "infrc:host=192.168.1.110,port=11114" -M r=0
Additional notes:
LogCabin (description of each step + the non-short version without using scripts):
0. setup
in logcabin/Examples/Reconfigure.cc:
- change the cluster address according to the group of machines i'm using (logcabin01, logcabin06, and so on)
- put the right addresses for the servers to be added
My current settings for various parts are (assuming i will be starting the logcabin instance first on rc10, followed by rc09 and rc08, in no particular order):
Cluster cluster("logcabin06:61023");
and
servers.emplace_back(2, "192.168.1.109:61023");
servers.emplace_back(3, "192.168.1.108:61023");
1. to forget initialization / data from previous run:
rm -r log
2. initializing logcabin (with first member on rc10):
./scripts/initlog.py --serverid=1 --address=192.168.1.110:61023
3. now start all the servers (including the first member with which i had initialized, and the ones that will be added during the reconfiguration) by doing this on each of them:
./build/LogCabin
4. to reconfigure with more servers:
make sure other servers i want to add are up and running (step above)
on some machine:
scons (to compile Reconfigure.cc)
./build/Examples/Reconfigure