Let's detect bots and prevent them from spawning WebSocketActors!

Sorry, failed to fetch structure.

[error] application - Could not get URL
org.webjars.MultipleMatchesException: Multiple matches found for d3.min.js. Please provide a more specific path, for example by including a version number.
at org.webjars.WebJarAssetLocator.getFullPath(WebJarAssetLocator.java:259)
at org.webjars.WebJarAssetLocator.getFullPath(WebJarAssetLocator.java:189)
at org.webjars.play.WebJarsUtil$$anonfun$locate$1.apply(WebJarsUtil.scala:48)
at org.webjars.play.WebJarsUtil$$anonfun$locate$1.apply(WebJarsUtil.scala:48)
at scala.util.Try$.apply(Try.scala:192)
at org.webjars.play.WebJarsUtil.locate(WebJarsUtil.scala:48)
at org.webjars.play.WebJarsUtil.url(WebJarsUtil.scala:125)
at org.webjars.play.WebJarsUtil.script(WebJarsUtil.scala:195)
at views.html.main$.apply(main.template.scala:88)
at controllers.Application$$anonfun$index$1$$anonfun$apply$5.apply(Application.scala:181)

Most conversions don't work (e.g. A3M to FASTA, FASTA to A3M, A3M to CLUSTAL). The main use of this tool is A3M to FASTA conversion.

We could implement a feature that jobs which are running get resubmitted after the Toolkit has restarted. This is quite easy to accomplish.

@vikramalva @davidmrau

there are some differences in a results json:

https://gist.github.com/zy4/8e8a2d88b0b9e00793258e62d52c3e87

I think it has nothing to do with mongodb.

The current image has a typo (sorry is spelled sory); could you please find us a new image?

The sweeping routine is time-related but the problem we want to address is space-related:
By setting a limit of days after which we delete jobs does not mean that we could not have exhausted our capacities before reaching this limit.
It would make more sense to wait with deletion until we run short of space.

... and the new alignment viewer has some issues with the charset (go to some dropdown menu)

I just noticed if you have a lot of jobs in the list it gets too long. Maybe we can limit it and give it a scroll bar from a certain height on.

We get quite a few error messages on shutdown (the mongodb collections and every connected websocket will throw one, for example)

@davidmrau do you have any idea why the trafficbar is gone again? It worked 3 days ago.

I think the version of the executable differs from rye.

TODO

Tries to offer jobs that are not existing anymore (see image), although there are some that ran through successfully. If one tries to load, the job manager page is shown, which essentially means the job is not found.

already reported to woerner & maan.

It is never used.

HHPRED_MANUAL displays the process log of some other currently running job.

Steps to recreate the error:

1. Start a new HHblits (or any other) job.
2. Go to this job: https://toolkit.tuebingen.mpg.de/#/jobs/2692785
3. Select a couple of hits and click on "Model using selection"; you should see the process log of the HHblits job.

@zy4 could you please take a look? HHpred results are forwarded to MODELLER a couple of hundred times a day and therefore we should try to fix this soon.

We have multiple validation processes running which decide on whether the button is disabled or not. We should have a FSM for buttons and validation responses in general written in client-side logic, maybe in scalajs.

https://toolkit.tuebingen.mpg.de/#/jobs/2012593

Way too big to load synchronously. Also a overflow-y is missing or not visible.

The job deletion routine writes the jobIDs of the deleted jobs into a file. The path for this file needs to be added to your Application.conf

In my case it is configured like this:

deletion_log_path="/ebio/abt1_share/toolkit_support1/drau/Toolkit/deletion.log"

After barely two weeks since deployment, the result collection has grown to the size of 4 GB. With this rate we will run into problems. We either should take care of sweeping old jobs soon or think about running a MongoDB sharded cluster. Beware: After the database stores more than 256 GB, you cannot start sharding anymore!

@JoKuebler can you optimize and compress it with https://pngquant.org/ ?
Thanks!

@vikramalva I have registered the FQDN

bioinformatics-toolkit.de

long ago. Should we use it or can I throw it away? If we want to use it, we need to either buy a certificate or create one on the server with Let's Encrypt (I do this all the time).

... when netty is down

e.g.
Start an HHpred job and when it is running, go to the 'parameters' tab and try to change 'No. of target sequences'

and reconnect button is gone

I don't know if this just came up for me because of some play2.6 upgrade related changes or if this error is not caught in master aswell.
It just continues loading while the server prints:
[info] application - Job was found but deleted.

Creating new jobs with the first user will fail as the host name is not set yet.
Possibly use this?
InetAddress.getLocalHost().getHostName()

There is still a huge mess from the deploying. Would be nice if you could clean this up a bit.

Sometimes after viewing a big resource where the spinner came up and after going somewhere else, e.g. the index page, the spinner still remains visible. This happens quite often so I think you already noticed.

... in order to detect malicious requests automatically

I have problems to scroll with 4 tabs open.

reconnecting-websocket.amd.js:1 Uncaught ReferenceError: define is not defined
at reconnecting-websocket.amd.js:1
(anonymous) @ reconnecting-websocket.amd.js:1

can somebody verify this?

I just got an email and the link of "Your Toolkit Team" refers to olt:1234. This should be changed soon.

In preparing for the workshop that I am going to be teaching next week, I realized that our Toolkit does not allow concurrent forwarding. If the students would try to forward results of a hashed job, only some are going to succeed, as the files generated for forwarding are written into the same job folder with the same name.

When I run a Marcoil job, the last three result tabs on the right are having content, when I move between them, the loading spinner appears and the content is all of sudden gone. This might also be the case for other tools.

1. for documentation
2. for the ease of deployment and maintenance