descartesresearch / teastore Goto Github PK

I was able to clean up the warnings across all projects, except for three warnings in the image provider.

The warnings concern some unused code in the TestRandomReplacement.java, i would rather not delete the code, in case it is still needed.

Additionally the SetupController.java contains two TODOs, either fix the TODOs or convert them to gitlab issues :-)

Inputting a username that does not exist leads to an error being thrown. In general, the load balancer throws a timeout exception if the returned status code is not 200.

usually found in catch blocks or servlets

A java.lang.IllegalStateException is thrown for every service because the web application instance has been stopped already. The state exception is followed by a javax.ws.rs.ProcessingException because the event listener cannot be destroyed.

If a user logs in using a wrong username or password he should get some kind of visual feedback on what happened. Maybe a red message bar?

make it fail gracefully if request comes in before registry is online

E.g. when getting an entity with an invalid ID.

If an image provider is started and no persistence is available it currently fails with the exception below. This is probably due to the recent changes, where a connection throws an exception if it takes too long instead of returning null.

tools.descartes.petsupplystore.registryclient.loadbalancers.LoadBalancerTimeoutException: Timout at endpoint: generatedb, with target service: tools.descartes.petsupplystore.persistence
        at tools.descartes.petsupplystore.registryclient.loadbalancers.ServiceLoadBalancer.loadBalanceRESTOperation(ServiceLoadBalancer.java:204)
        at tools.descartes.petsupplystore.registryclient.loadbalancers.ServiceLoadBalancer.loadBalanceRESTOperation(ServiceLoadBalancer.java:177)
        at tools.descartes.petsupplystore.image.setup.SetupController.waitForPersistence(SetupController.java:120)
        at tools.descartes.petsupplystore.image.setup.SetupController.fetchCategories(SetupController.java:172)
        at tools.descartes.petsupplystore.image.setup.SetupController.generateImages(SetupController.java:218)
        at tools.descartes.petsupplystore.image.setup.SetupController.startup(SetupController.java:542)
        at tools.descartes.petsupplystore.image.setup.ImageProviderStartup$1.callback(ImageProviderStartup.java:57)
        at tools.descartes.petsupplystore.registryclient.RegistryClient$1.run(RegistryClient.java:168)
        at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
        at java.util.concurrent.FutureTask.run(FutureTask.java:266)
        at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$201(ScheduledThreadPoolExecutor.java:180)
        at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:293)
        at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
        at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
        at java.lang.Thread.run(Thread.java:748)

tools.descartes.petstore -> tools.descartes.petsupplystore

don't forget to change poms, Service, web.xml, Eclipse context paths, etc.

Foreign key exceptions if a database larger than the previous is created (sometimes)

• remove autorun for startup daemon
• make it callable

It should also make a little more sense

Startup times on existing images are slow. There may be a syntax error in the base dockerfiles's ENV entry.

See title, we need more detailed monitoring

should also clean on startup after unexpected shutdwon

Specifically what it seems to do is the following:

• Sort products by number of orders in descending order
• For each number of orders (in descending order): select the first product with this order count and add to recommender output

If attached a small CSV sheet that shows the behavior and order counts for the default DB
recommenderanalysis.txt

Currently, if the user manages to somehow get a corrupted session cookie i.e. a cookie containing null, the user gets permanently locked out. We should be a little more robust here.

Using JavaEE session API will not work with distributed WebUIs

The image provider's startup could be refactored to be more easily testable. It would also help in finding errors and understanding the code base. Maybe some sort of state-machine?

back to shop button

Use the c:url tag. For an example, see footer.jsp

Commented them out, they concern calls to recommend before training.

This used to work. It's broken when starting docker containers.

Reseting the database must reset all:

• Persistence providers
• Image providers
• Recommender services

Eclipselink throws an error

Recommender is the last to include Persistence and Registry imports during testing. The imports are configured wrong, though. Migrating to Wiremock should fix the issue.

• Configure SLF4J to a reasonable setting.
• Initialize logger in registry client
• Replace all Sysout/syserr with Logger calls (with useful log levels)

Occurs sporadically with a null pointer in the registryclient callback.

It's too large for mobile

The single container section is still empty