Problem with Strings to Document 5.4.2

KNIME Analytics Platform
1 
Hi everyone, I'm having a small problem. When I try to use the Strings to Document node in KNIME version 5.4.2, I get this error:

"Could not store document in cell: PAGOBANCOMAT MAESTRO CA230
ERROR
Strings to Document 9:2495 Execute failed: Cell at index 0 is null!"

The node with the same configuration in version 4.6 (I know it's a bit dated) works correctly. I've attached a screenshot of the two configurations.

4.6

4.6
4.61807Ă—748 95.5 KB


5.4.2

5.4.2
5.4.21851Ă—773 98.8 KB

Thank you so much for your help

2

Hi @cridiaz ,
Thanks for reporting the issue.
Could you please also share the complete log file, along with the KNIME workflow and dummy data(if the actual data is confidential)? This would help us reproduce the issue on our end.

Thanks,
Sanket

1 Like
3 
Hi, thanks for the reply. I'll test it because the problem might not be on the node but in our environment.
Anyway, as soon as it occurs, I'll share the complete log with the problem.
4

Hi, unfortunately I don’t have any further information. The only thing the console and the log give me is this.

ERROR  KNIME-Worker-2 DocumentBufferedFileStoreDataCellFactory  Could not store document in cell

ERROR  KNIME-Worker-9-Dictionary Tagger 3:255:214 Node  Execute failed: Cell at index 0 is null!java.lang.NullPointerException: Cell at index 0 is null!

the strange thing is that it happened on another node the “Dictionary Tagger” and not on the String to Document

5

Can anyone help me? Now the error occurs even with version 4.6 and it never did before

6

Hi @cridiaz -

Can you upload a small workflow with dummy data that reproduces the problem? That would help a lot.

2 Likes
7

Hi @ScottF I’ll try to attach a portion of the WKF file, reading a .table file and two processing nodes. As mentioned, it could also be a problem with our data.
Thanks for your help.

DATI.xlsx (484.4 KB)

WKF_STRING_TO_DOCUMENT.knwf (18.7 KB)

8

Sorry, I’m not able to reproduce the problem in KNIME AP 5.9. But I also notice the workflow and data you uploaded don’t seem to correspond to your previous screenshots, so I’m not sure where the discrepancy lies.

If I had to guess, I’d say there’s an issue with your data encoding that’s throwing things off? But the data you provided works for me…

9

Hi, unfortunately, due to size constraints, I only uploaded a portion of the data. In total, it has to process about 3,000,000 rows. Like you said, this is probably a problem in our environment, because sometimes, after trying again a few minutes, it works correctly, making it difficult to reproduce the error. Thanks again for your help. If I can reproduce the error with the debug log, I’ll post more information.
I forgot, the processes that are having the problem are two different ones: one is the screens one and the other is the one I posted.
Thanks again.

10

@cridiaz could you think about doing it in chunks and maybe record which of them were successful and redo them if it only occurs occasionally.

https://medium.com/low-code-for-advanced-data-science/knime-snippets-1-collect-and-restore-or-how-to-handle-many-large-files-and-resume-loops-c57795b65d7e

Also you sure have checked to give KNIMR enough RAM.

11

A solution might be to do it in blocks; I hadn’t thought of that. RAM could actually be a possible cause of the problem, because we had some problems a while ago and then it started to malfunction. I’m trying the block solution because I can’t increase the RAM.

12

You should really try and see if you can increase the RAM because the default setting is not that large.

13

I have already set the Xmx parameter to 20g in the knime.ini, unfortunately I can’t set it higher because the machine doesn’t have more, we noticed that when launched at different times where the machine could be more discharged it works without problems

14

My experience is that you should leave some for the system. So maybe only go as high as 2/3 of your RAM. So maybe 15 or 16 max. Otherwise I have also seen problems.

This seems to be an indicator that indeed it might be affected by memory problems. So maybe try chunks or even streaming.

15

Thanks @mlauber71
I’ll try streaming and see if the error goes away. Also, one of the two processes mentioned is scheduled and usually runs between 11 PM and 2 AM. We’ve noticed that it works perfectly at a different time, so 99% of the time it’s a RAM issue, as other processes are running during that time.

Thanks and happy holidays.

1 Like