The ShiftML tool does not seem to be working

[brusjiri]

Hello everyone

I am having a problem with the ShiftML tool on the website: https://www.materialscloud.org/work/tools/shiftml. I have the feeling that the tool is not working. I see no activity after uploading the cif file or starting the model calculation. Can you give me some advice on how to start calculating the chemical shielding parameters? Many thanks. Jiri Brus

[eimrek]

thanks @brusjiri. This should be fixed now. Feel free to reopen if it comes up again.

[jakub126746]

Hello @eimrek , this tool doesn't work for me either. I tried to calculate shifts for my own structure and also for an example structure and in both cases I got the following error in the console: Is there a possibility to fix this issue? Thank you.

[eimrek]

thanks for the note @jakub126746

Indeed, backend of the tool seems down again. I restarted it and it's working now, but i suspect it can stop again in the near future.

As this is a contributed tool, we (the materials cloud team) don't have expertise (nor time currently) to investigate in detail, to fully maintain and fix it.

@Luthaf do you have ideas on who could help with this? On a separate note, it would be great if this tool could be migrated to the dokku platform (but for that it needs to be using a single docker container).

[Luthaf]

I don't have much experience with ShiftML itself, it was written before my time! @ceriottm do you know who wrote this code & who can help to fix it?

I'm happy to help moving it to dokku, depending on the current architecture it could be a lot of work or just an afternoon.

[ceriottm]

Seems to be working right now. I can't see why it would become suddenly so unstable. In all events we'd like to make a new better version so perhaps it's best not to invest the time to port this. We'll make plans in the coming 2-3 weeks I think

[eimrek]

thanks @Luthaf, @ceriottm. Could be that a specific input (maybe too large, or with some error) crashes the backend and a user has tried it recently multiple times?

Either way, I can just restart the tool when i get notified that it's not working. Great that you're planning a new version.

[ceriottm]

thanks @Luthaf, @ceriottm. Could be that a specific input (maybe too large, or with some error) crashes the backend and a user has tried it recently multiple times?

Maybe, IDK if you have logs that might help in that respect. If that the case we might be able to quickly set a limit on number of atoms.

[brusjiri]

Hello everyone,

Thank you very much. I just tested it and it really works. Everything seems to be OK.

Thank you so much, this is great. JB

--

Jiri Brus, PhD Senior Research Fellow / Department of Structure Analysis Institute of Macromolecular Chemistry, Czech Academy of Sciences tel.: +420 296 809 350 e-mail: @.*** Heyrovského nám. 2, Prague 6, Czech Republic www.imc.cas.cz(https://imc.cas.cz) (https://www.imc.cas.cz/en/)   The contents of the report do not constitute a proposal or acceptance of a proposal to conclude or modify a contract. Contracts shall be concluded exclusively in writing and must be signed by the legal representative – the Director of the Institute.

---------- Původní e-mail ---------- Od: Michele Ceriotti @.> Komu: materialscloud-org/issues @.> Kopie: brusjiri @.>, Mention @.> Datum: 23. 4. 2024 22:29:23 Předmět: Re: [materialscloud-org/issues] The ShiftML tool does not seem to be working (Issue #34) "

Seems to be working right now. I can't see why it would become suddenly so unstable. In all events we'd like to make a new better version so perhaps it's best not to invest the time to port this. We'll make plans in the coming 2-3 weeks I think

— Reply to this email directly, view it on GitHub (https://github.com/materialscloud-org/issues/issues/34#issuecomment-2073388082) , or unsubscribe (https://github.com/notifications/unsubscribe-auth/BHRAONAOSCYURDEO63IO6OTY63ACHAVCNFSM6AAAAABFUVQ5EWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDANZTGM4DQMBYGI) . You are receiving this because you were mentioned. Message ID: < @.***> "

[NMR-GitHub]

Greetings,

problem with calculating shifts remains; clicking on "calculate shifts" does not do anything, not even with an example structure. Could you repair it, please? I am student and I use this tool for my studies, thank you.

ES

[eimrek]

hi @NMR-GitHub, thanks for the notice. I restarted the app and now it's working.

I also checked a bit: the redis docker container exited with a status code of 255 on 19th of June.

Here's are the outputs of docker logs and docker inspect commands on the exited container:

docker-logs.txt docker-inspect.txt

Perhaps these could be useful to understand what causes the issue. @Luthaf @ceriottm

From a quick glance, it doesn't seem to be an OOM error. Also the logs don't show anything around 19th of June.

[NMR-GitHub]

Thank you, it works now.

[NMR-GitHub]

Hello, it doesn't work again. Is there something I could perform to make it work? Thank you.

[eimrek]

Thanks for the note @NMR-GitHub, just restarted the service, should be working now.

Also pinging @bananenpampe, so you're aware. I think you're looking into this issue, right? The redis container was crashed again.

[bananenpampe]

Thank you @eimrek for letting me know. I had a look at the container logs you had sent me last time, and I was not able to identify any messages that allowed my to further identify what the issue was

[eimrek]

ok, i took another look and it seems that the redis contiainer just receives SIGTERM for whatever reason:

1:signal-handler (1720602229) Received SIGTERM scheduling shutdown...
1:M 10 Jul 09:03:49.426 # User requested shutdown...
1:M 10 Jul 09:03:49.426 # Redis is now ready to exit, bye bye...

Also, the container didn't have a restart: always policy, which i added now. This should make it that it automatically just restarts when this happens again. This hopefully solves this from the users perspective.

Closing the issue. Please open it again if this resurfaces.