How to retrieve processes from an online repository...

MacPhotoBiker

Hi,

I was using remote repositories in a local RapidMiner installation. I was also using RapidAnalytics to create online charts, and all worked amazingly well. Unfortunately, all over sudden it stopped working, and I get the error message that RapidMiner can't connect to RapidMiner Server repository (I'm using RM 5.3.015, I was not upgrading to version 6 or so), I also made no update to RapidAnalytics!

Would there be ANY way for me to retrieve the processes that were stored online, so I can keep using them locally?

I have put really significant efforts in designing processes, and not being able to access them now is a quite shocking experience  

I don't know (yet) what lead to this situation, and it doesn't really matter. But I'd REALLY appreciate some help in retrieving these processes! There are hundreds of hours of work in them....

Thank you VERY much for any help!!

Marco_Boeck

Hi,

first things first: Don't panic

Can you still access the RapidAnalytics web interface? If so, you can download the repository as a zip file in the Repository Browser (Download section on the right side).

If not, we have to fix your RapidAnalytics installation.
My answer assumes you are using a RapidAnalytics instance on a machine you have access to:

1) To be safe, create a complete database dump of the DB RapidAnalytics was using.
2) Is the server still running? If so, stop it.
3) Try to restart it and see if you can access it now.
4) Check the server.log in the SERVER/default/logs folder. You should see some errors there. Maybe you can figure out the problem by reading the error messages. If not, post them here.

Alternatively, you could try and create a separate, fresh installation of RapidAnalytics and point it to the existing database. Whether this works or not depends on the integrity of the database.

Bottom line is, as long as you have a dump of the (non-corrupted) RapidAnalytics database, your processes should be salvageable

Regards,
Marco

MacPhotoBiker

Hi Marco,

thank you very much for your prompt reply! And thank you for pointing out the database! I knew my data was stored in the database, but somehow I didn't think of the processes also being stored there, instead I was digging through the server folders, looking for some xml files that I assumed would hold the processes. I got a local copy of the database now, and I'm pretty sure it's there's nothing wrong with the database itself. Phhhheeeewwwww.....


I had tried to reboot the entire (VPS) server, but in vain, so I looked at the log file you mentioned. There's indeed a message, but I'm not quite sure what to do with that.

These are the first few lines of the log file:

*********************************************************************************************************************************************************************************************************************

2014-02-02 23:56:12,012 INFO  [com.arjuna.ats.arjuna] (Transaction Expired Entry Monitor) ARJUNA-12296 ExpiredEntryMonitor running at Sun, 2 Feb 2014 23:56:12
2014-02-03 11:56:12,035 INFO  [com.arjuna.ats.arjuna] (Transaction Expired Entry Monitor) ARJUNA-12296 ExpiredEntryMonitor running at Mon, 3 Feb 2014 11:56:12
2014-02-03 14:19:37,022 ERROR [ProfileServiceBootstrap] (Thread-2) Failed to load profile:: org.jboss.deployers.client.spi.IncompleteDeploymentException: Summary of incomplete deployments (SEE PREVIOUS ERRORS FOR DETAILS):

DEPLOYMENTS MISSING DEPENDENCIES:
  Deployment "MK2TimerServiceFactory" is missing the following dependencies:
    Dependency "persistence.unit:unitName=jboss-ejb3-timerservice-mk2.jar#timerdb" (should be in state "Installed", but is actually in state "Configured")
  Deployment "jboss-switchboard:appName=RapidAnalytics-CE-1.3.013,module=RapidAnalytics_EJB,name=MigrationEJBImpl" is missing the following dependencies:
    Dependency "jboss.j2ee:ear=RapidAnalytics-CE-1.3.013.ear,jar=RapidAnalytics_EJB.jar,name=UserManagementEJBImpl,service=EJB3" (should be in state "Installed", but is actually in state "PreInstall")
    Dependency "jboss.j2ee:ear=RapidAnalytics-CE-1.3.013.ear,jar=RapidAnalytics_EJB.jar,name=RapidMinerEJBImpl,service=EJB3" (should be in state "Installed", but is actually in state "PreInstall")
 

*********************************************************************************************************************************************************************************************************************

Thank you very much again for looking into it. I will now try to build a local installation, but I'd still like to try and get the server installation running, too.

Best regards

MacPhotoBiker

MacPhotoBiker

Just to add another piece of information, this is the error message I see when I try to access RM:

http://i.imgur.com/qa4ANsc.png

Thank you!

Marco_Boeck

Hi,

your server fails to deploy RapidAnalytics for some reason and the image does not contain enough information to figure out why. You have two options right now:

Option 1:
1) Shutdown the existing RapidAnalytics (aka the JBoss server it runs on via shutdown.bat/sh).
2) Install RapidAnalytics into a different folder (using the same settings and pointing it to the exact same database [of which you got a backup now]) and try to start the new installation. If the database itself does not contain broken entries, it will start and work like before. You are done at this point.

Option 2:
1) Shutdown the server if it is still running (aka the JBoss powering RapidAnalytics)
2) Delete the boot.log and server.log files
3) Try starting it again, post (or link) the complete boot.log and server.log here so we can find out what's wrong and try to fix the problem


Regards,
Marco

MacPhotoBiker

Hi Marco,

thank you very much again! Thanks to your advice I was able to restore the database locally. I also installed RapidAnalytics locally, and I am able to browse the Repository as well as the services which I had exported. Everything is back, that is amazing, you just saved my weekend. THANK YOU!

Before I touch anything on the server, I'd like to make sure I get everything running well on my local machine.

The last thing I could not yet figure out is this:

How can I tell RapidMINER (5.3.015) to connect to the local repository?

Thanks again for helping me out with the "Disaster Recovery"

Gert

MacPhotoBiker

Hi Marco,

I just figured out how to connect to the repository again.

My error was that I tried to configure a local repository. However, I'm running RM on a local server, so I figured that for RA this is still a Remote Repository. All I had to do is to provide the repository URL (http://localhost:8080), and user credentials for Rapid Miner.

All I need to do now is to rewire the database connections - or I might even leave them as is, running RA and RM on a local server, but still writing the results to an online database.

Thank you very much again for all your help, I got everything back and just need to do some minor tweaks now.

MacPhotoBiker

Hi,

sorry for opening this ticket again.

As described above, thanks to Marco's help I was able to transfer the RapidAnalytics database from a online webserver to my local machine (still using an Apache server), and I did not lose a single process. But now I ran into an issue:

In RapidMiner, I can create new processes, and store them in my existing RapidAnalytics repository. Then, I can go to RapidAnalytics, "Browse repository", and I find the process I just stored. But then, when I try to run the process, I get an error message that the repository does not exist:

http://i.imgur.com/TLMzsGo.png

I went through all settings, but I have not been able to sort this out. I'm using  RM 5.3.013 and RA 1.3.013.

Thank you very much for any help!

Marco_Boeck

Hi,

your process retrieves repository data via an absolute path. The absolute path however contains the name of the repository which is something only your local RapidMiner instance knows and can resolve. RapidAnalytics will fail to do so with the error message you posted. This is also displayed in the "Problems" view which should show a warning stating that a repository is accessed by name which is not portable. You have to use relative paths when you want to execute processes on RapidAnalytics. To go up folders, you can use

../

for each parent folder you want to acess.

Regards,
Marco