doc-vu / pads Goto Github PK

assets directory not found:

 [gme:server:standalone] Requested DecoratorSVG not found /assets/DecoratorSVG/Settings-toolkit.svg
2017-12-02T17:09:54.883Z - warn: [gme:server:standalone] Requested DecoratorSVG not found /assets/DecoratorSVG/png/follower.svg
2017-12-02T17:09:54.904Z - warn: [gme:server:standalone] Requested DecoratorSVG not found /assets/DecoratorSVG/png/leader.svg
2017-12-02T17:09:54.922Z - warn: [gme:server:standalone] Requested DecoratorSVG not found /assets/DecoratorSVG/png/learner.svg

Link Directory is configured at https://github.com/doc-vu/pads/blob/master/config/config.default.js#L25

we could integrate code editor so that the code that is generated is viewable in the panel itself

need to create tutorial examples, documentation for the MODELS tutorial.

For now, the pads will simply generate a Mininet topology and a list of commands. The user may type those commands in the Mininet Client in order to run their applications. This is fine when the experiment is in relatively small scale. But when there are a large number of applications that need to be deployed on the Mininet, it will be a huge issue that the user has to type a command every time they want to run an application. For example, in a publisher-subscriber-broker system, if we have one publisher running on host1, and one broker running host2, and 20 subscribers running on host3. The user may expect the pads also generate a script that automatically runs those applications. I wrote a code snippet for this purpose:

def runApps(net):
    "Run cmds in command.txt"
    print "Running applications"
    with open("command.txt") as f:
        for line in f:
            splitedLine = line.split(' ', 1)
            print splitedLine[0]
            host = net.get(splitedLine[0])
            print splitedLine[1]
            host.cmd(splitedLine[1])

and our generated commands.txt should look like this:

h0 ./runBroker.sh &> ./result/h0broker0.log
h1 ./runPub.sh 10.0.0.1 &> ./result/h1pub0.log
h2 ./runSub.sh 10.0.0.1 &> ./result/h2sub0.log
h2 ./runSub.sh 10.0.0.1 &> ./result/h2sub1.log
h2 ./runSub.sh 10.0.0.1 &> ./result/h2sub2.log
...
...

Notice that every command has an redirect command as a part of it. It's reasonable that every application running on the Mininet need a private place to save their output. We could use "hostName" + "instanceID" to make the logfile for every application unique. And user should not bother to pass a unique logfile name as one of the arguments for each of applications.

currently by default needs an ssh key. make that configurable

add preliminary support for constraint checking