@kumarsen26 Being a vertically integrated stack, SwimOS comes with a built-in database that tracks the state of all lanes that are not explicitly declared transient, but only if you opt into persistence. The easiest way to do this is via the recon configuration file; see https://github.com/swimos/traffic/blob/master/server/src/main/resources/server.recon#L14-L16 for an example. Note that the directory to which the database files are written is configurable.
Upon restart, Swim will load the same values into the lanes that they had immediately before the shutdown.
@brohitbrose yes WebAgent is started verified it by
@Override
public void didStart() {
logMessage("Web Agent Started");
}
Also created one lane
@SwimLane("hello")
ValueLane<String> info = this.<String>valueLane()
.didSet((newValue, oldValue) -> {
logMessage("info set to " + newValue + " from " + oldValue);
});
Both are non empty. Also pushed some data into other lanes.
But, Still no store generated and on restart other lanes products are empty
@kumarsen26 I think I've got it. You're likely missing a module dependency. Please make sure that you have the following (or something similar to it) in the dependencies { ... } section of build.gradle:api group: 'org.swimos', name: 'swim-store-db, version: '3.10.2'
Furthermore, if you're using Java 9+ with modules, you'll need the following in your module-info.java:requires swim.store.db;
server.recon file. Unfortunately in Heroku you don't have control over which port is exposed on the server since the port is assigned dynamically. Heroku does expose this port as an environment variable, so you'll have to read this property and inject that into the server configuration.
I am referring https://github.com/swimos/greenhouse/tree/feature/docker
But not able to find persistance example with docker.
- Remove the line
ServerLoader.loadServer();in your main class. Instead use the CustomServerLoader provided here: https://gist.github.com/ajay-gov/282ac4c28ed5027b0cecb75eb3f7c918. So your code then would beCustomreServerLoader.loadServer()(Note: You need to change the package name of the CustomServerLoader to match your package name. - Not the line here: https://gist.github.com/ajay-gov/282ac4c28ed5027b0cecb75eb3f7c918#file-customserverloader-java-L113, this is the part which reads the system property called
portand uses the value in that property and gets the swim application to bind to that port- You need to pass the System property to the gradle run task. Unfortunately gradle doesn't do this by default. So add the following line to the gradle run task: (using this as an example gradle file: https://github.com/swimos/cookbook/blob/master/planes/build.gradle#L59)
systemProperties System.getProperties(). So the run task would look like this:run { dependsOn jar systemProperties System.getProperties() doFirst { jvmArgs += [ '--module-path', files(configurations.runtimeClasspath, jar.archivePath).asPath, '--module', "${moduleName}/${mainClassName}" ] classpath = files() } }
- You need to pass the System property to the gradle run task. Unfortunately gradle doesn't do this by default. So add the following line to the gradle run task: (using this as an example gradle file: https://github.com/swimos/cookbook/blob/master/planes/build.gradle#L59)
- Override the gradle run task that heroku invokes by doing the following:
heroku config:set GRADLE_TASK="run -Dport=$PORT". If this is not how you run then modify the run task suitably so that the-Dportvariable is assigned to the$PORTvalue
server.recon file to use that mount point. So if your mount point of the docker persistent volume is /mnt/docker/v1 then change the store path in the server.recon file as path: "/mnt/docker/v1/swim-test/"
9 replies
Hi,
Could you please tell me how to get particular values from a Maplane (ie I am expecting only some values based on query params like http get method)
Currently Plane and Agent is like below
in Plane
@SwimRoute("/customer/:name")
private AgentRoute<CustomerAgent> customer;
in Agent
@SwimLane("view")
Guidance is appreciated.
You may refer to this: https://www.swimos.org/tutorials/map-lanes/ and scroll to the section that says “Reading from Map Lanes”.
After opening a downlink to the lane, you can simply call get() on the key for a one-time read. To receive streaming updates instead, you can utilize the didUpdate(), callback, and ignore values for which the key doesn’t match your target.
@brohitbrose Thank you !!!
If i have one client and multiple server applications.. How do we share the data since it is stateful?
How to load balance?
is this needed? or it will be only one client and one server?
I am just thinking in prodution perspective and it will be good some architecture is shared..
sorry if i missed if it is available.
@kumarsen26
While we don't have a full-fledged tutorial for what you're looking for, it follows the process here: https://github.com/swimos/transit#run-as-a-fabric
The main idea is that Web Agents can be distributed over multiple machines, yet each machine can serve data from Web Agents that run on a different machine (e.g. downlinking to a Web Agent that runs on machine2 will work, even if you set your hostUri=machine1). This works because Swim servers automatically abstract over the load balancing logic that you described.
You'll have to seed each server is seeded with the right recon file. I recommend reading the run scripts and the .recon configuration files, then asking us any questions if anything is unclear.
Hi,
In client, I am getting below response in client side when trying to read value from mapLane.
What I am getting.. (@Customer and @ContactInfo is the object name)
{
"@Customer": null,
"firstName": "kumar",
"lastName": "Raju",
"dob": "1990-09-09",
"contactInfo": [
{
"@ContactInfo": null,
"email": "test@gmail.com",
"phone": "1252425262"
}
]
}Expected is
{
"firstName": "kumar",
"lastName": "Raju",
"dob": "1990-09-09",
"contactInfo": [
{
"email": "test@gmail.com",
"phone": "1252425262"
}
]
}My Client code:
swim.downlinkMap().hostUri(HOST).nodeUri('/customer').laneUri('details').open().didUpdate((key,value) => {
console.log({...value.toAny()});})I am not sure what I am missing.. Could you point the mistake.
Hard to say for sure without seeing the server code. But you can skip to past the --- for my suggested fix.
My hypothesis is that this is actually the expected behavior. I assume that
- You have a Customer Java object with
firstName,lastName,dob, andcontactInfo - The type of
contactInfoisContactInfo[]
Then the recon (not JSON) serialization of your object will look like:@Customer{firstName:kumar,lastName:Raju,dob:"1990-09-09",contactInfo:{@ContactInfo{email:"test@gmail.com",phone:"1252425262"}}}, assuming that you used annotations instead of manually writing the Form<Customer>.
When a serialization looks like @foo{bar:a,baz:b}, the swim.structure representation of this is actually a three-length Record (i.e. a collection of Items) whose first item is an Attr with key foo and undefined value; second Item is a Slot with key bar and value a; last Item is a Slot with key baz and value b. It is by design that an Attr appears "raised" even though it is in the same structural level. JSON does not differentiate between "special" key-value pairs like Attr vs Slot, so when you translate an Attr-containing object to JSON, it simply appears as a "regular" key-value pair with @ in it.
---
To get everything except for the first Item in a Record, you should be able to call tail() on it. On the shorter example above, you should see (in JSON) {bar:a,baz:b}. In your example, you would need to do this twice to handle the nested ContactInfo type object.
@brohitbrose
Please find server code.. I hope this helps..
Plane:
@SwimRoute("/customer")
private AgentRoute<Customer> customer;Lane:
@SwimLane("details")
MapLane<String, Customer> view = this.<String, Customer>mapLane().didUpdate((key, newValue, oldValue) -> {
System.out.println("the new value is " + newValue.toString());
});
customerDetails and contactInfoDetails (this might take a little more processing, maybe something like contactInfoDetails.getItem(0).tail() for the first value since it's an Array)as defined above
- How to scale out ? Can I run a dynamic cluster ?
- how to backup and administrate ? can I store a file as well ?
- Can I just run the Java server then use a nodejs server to handle logic ? or I need to program everything in Java for the server ?
- How the cache and security works ? should let the client contact swim server directly or through proxy ?
- How about authorize and access control ?
@meepeek Good questions.
- You can run swim in a cluster. Please look at this example of the transit application in a 2 node cluster: https://github.com/swimos/transit#run-as-a-fabric
- All swim lanes can be persisted if you turn the persistence option on. This is part of our enterprise offering. If you turn the persistence option on, then the data stored in the lanes for all agents of a specific type will get written into files on disk which you could then back up.
- Ideally it's best to program all the server side logic in java using the SwimOS platform
- The client can connect to the swim server directly. Web agents expose endpoints and streaming APIs, so data from the lanes will get streamed to your client
- You can configure access control and authorization programmatically. This typically done by injecting an Authenticator and a Policy to the Swim Plane. Here is an example from the cellular app: https://github.com/swimos/swim-cellular/blob/master/src/main/java/swim/cellular/CellularPlane.java#L24-L30
Please do let us know if you have any other questions. Also one question for you, how did you hear about the Swim platform?