Containerized version

[mattjmeier]

Hello,

I was wondering if you have any plans to release a containerized version of runscanner. Since MISO LIMS runs as a docker stack by default, it seems like it would be easier to integrate into the environment that way.

[djcooke]

We don't have plans to do this, but I agree that it would be ideal. We'll keep this issue open as a request

[mattjmeier]

Thanks!

In the mean time, if I attempt to make a dockerfile, do you have any suggestions about what image I should use as a base? If I get anywhere with it I will share it with you.

[djcooke]

That would be appreciated! tomcat:9-jdk11-openjdk-slim should be a good base - it's the same that the MISO webapp uses

[mattjmeier]

Hello,

I am putting something together here. I don't really have much experience building containers from scratch though. Probably better ways to do things... especially the Maven installation (just found an example online that I copied).

I'm not sure that it works yet though. The web server is running, and I have runs mounted to the /srv/sequencer folder. The problem is that it doesn't see existing runs... is it supposed to? I couldn't really tell from the documentation. When I run java -cp $RUN_SCANNER_HOME/WEB-INF/classes:$RUN_SCANNER_HOME/WEB-INF/lib/'*' -DplatformType=ILLUMINA -Dname=default -Dtz=America/Toronto -Dparameters={} ca.on.oicr.gsi.runscanner.scanner.FindRuns /srv/sequencer/NextSeq500/ it does indeed output run folder names, so I'm not sure where to go from there. I speculated that it could be a permissions issue but setting GID and UID for the container did not help. Here is what the site shows currently (and there is nothing listed in the other tabs):

Core

Uptime	PT27M32.207114S
Start Time	2022-02-18T18:52:04.485158Z
Run Scanner Version	1.13.2-SNAPSHOT
Is Configuration Good?	Yes
Last Configuration Read	2022-02-18T19:08:04.346811Z
Scanning Enabled	Yes
Currently Scanning	No
Time Since Last Scan	PT11M32.341652S
Processed Runs	0
Waiting Runs	0

/srv/sequencer/NextSeq500

Maybe I need to run a command to force import existing runs..? Let me know if that is the case...

Anyway, this could probably easily be added to the docker-compose for MISO LIMS but I haven't got that far yet. I suppose it might even be possible to build the Run Scanner software on top of that base image, but I don't really understand the staged build concept yet.

Thanks!

[mattjmeier]

As a follow up, I can also confirm that java -cp $RUN_SCANNER_HOME/WEB-INF/classes:$RUN_SCANNER_HOME/WEB-INF/lib/'*' -DplatformType=ILLUMINA -Dname=default -Dtz=America/Toronto -Dparameters={} ca.on.oicr.gsi.runscanner.scanner.ProcessRun /srv/sequencer/NextSeq500/${run_name} returns a huge JSON result with the expected results... so I am guessing I just need to somehow tell it to look for existing runs?

[djcooke]

The name in runscanner.json is meant to refer to one of the processors in Run Scanner. It should be "default" for all Illumina sequencers. Besides that, I think everything looks good.

It's probably a bug that the page shows config is good when an invalid processor was specified, so I created separate issue #133 to investigate that

I'm very curious... Is "Itchy" the name of your sequencer? 😆

[mattjmeier]

Ohhhhh haha. That's hilarious - yes, it's importing runs now that I changed it back to default. Thanks very much for the help!

Yes, we have two NextSeq 500s... Itchy, and of course, Scratchy.

[djcooke]

I was hoping there was a Scratchy to go with it! 💯

[mattjmeier]

One more thing I will need to do is figure out how to make the database persistent for this. Any ideas what I need to mount to make that work? Should I actually be running a database as a separate service?

[djcooke]

Run Scanner doesn't use a database. It holds the runs in memory and has to re-scan them all after a restart

[mattjmeier]

Oh, that makes sense. I did notice that it works pretty fast (we have less than a thousand runs in total) so obviously not a big deal to rescan.

Anyway, what I've put together in that repo seems to work at my end and communicate properly with the MISO LIMS detailed demo stack. I noticed that I had to actually add both instruments using their Illumina identifiers to actually get MISO to import the runs - it might be worth mentioning this in the documentation somewhere, like here (perhaps as a Step 5 under Deploying), since I didn't see that information explicitly laid out.

Feel free to close the issue if you're happy with the example Dockerfile I've posted. Use whatever you'd like from that repo obviously... if it's something that you guys want to pursue.

The way I am running it is basically as a separate stack right now, but like I suggested I'm sure it would be easy to integrate into the other compose file - then you wouldn't need to expose ports on the host machine since it could communicate in the docker overlay networks.

All the best, and thanks again for your help!