Question: Bowtie 2 An error occurred setting the metadata for this dataset
0
lukacdm • 0 wrote:
Hi: the message "An error occurred setting the metadata for this dataset" is appearing in all of my Bowtie2 runs today, regardless of server. The auto detect button is not correcting the error. Thanks.
ADD COMMENT
• link
•
modified 12 months ago
by
Jennifer Hillman Jackson ♦ 25k
•
written
13 months ago by
lukacdm • 0
Where are you working? https://usegalaxy.org or elsewhere?
And by "server" do you mean the target cluster that the job is running on (job resources specified)?
Hi I am experiencing the same error with my analyses with bowtie2 too usegalaxy.org, default job resource)! I saw this could be related to some operations that have been made (don't know exactly what...) in a similar post about tophat. A proposal is reported there to resolve this by changing job resource option at the top of the tool form. I'm trying, I'll report the result soon. Any other suggestion is welcome however, if needed... thanks a lot
Olivier
hi - Thanks for letting us know. I can see this is occurring for jobs that are run on at least two different clusters. TACC beta should be avoided, but the default job resources shouldn't be producing datasets in this state (your job happened to be run on Jetstream).
I've asked our admin to look at this. For now, the solution is to use the link inside the expanded dataset to reset the metadata. Do this before running any downstream tools on datasets that end in this state.
Thanks for reporting the problem. I'm going to leave this open until resolved.
Hello Jennifer,
thanks for your reply. I see it's not a trivial point to solve on your side, we'll be patient, sure. I said in my post I'll report the output result of my test... didn't work, even changing resource as you mentioned in tophat problem post. Just one question: where is the reset metadata link you tell us about? Is it the yellow box asking to set manually or retry auto-detection? It doesn't work for me.
Olivier
Yes, that is the link. And if it isn't working... that's another problem we'll need to look at.
Two potential workarounds:
1 - Another way to reset metadata is to click on the pencil icon for the dataset to reach the Edit Attributes forms. On the first tab is a button to autodetect metadata - does that work or also fail?
2 - I imported a copy of your history and the metadata issue resolved. What happens if you make a copy of the history yourself (History menu > Copy History)? Does that fix the problem?
The prior issue about Tophat was different. It was cluster-specific, which is why changing the job resources helped (until that problematic cluster had a fix implemented). We don't know what is wrong yet with this issue. Not all Bowtie2 jobs end this way so it is difficult to reproduce. Please keep a copy of your original history if possible - or even just these jobs (and the inputs) to reduce how much of your quota is used up.
We'll get back to you soon, likely tomorrow, with at least an update and ideally a fix.
Thanks again for all the info and feedback - helps a great deal.
The first solution throw the following error: This dataset is currently being used as input or output. You cannot change metadata until the jobs have completed or you have canceled them.
The second seems to work! You've pointed it right. I could launch my paused operations and it didn't complain, so I think it's ok. Ok that all for now. Thank you again for those tips!
Olivier
Ok, glad that worked. And I've been able to reproduce the issue for some jobs.
If it makes it easier, using "Copy Dataset" and just putting it back into the same history also worked for me.
I'll make this in the wait for a correction if I needed, ok. I had to restart the following task (rerun a new task) of my workflow however, because it didn't resume despite it told me it was ok to resume. Does this kind of error break workflow irreversibly? I couldn't find the option specifying to resume the next jobs of the worklow in the bowtie2 parameters sheet? (I used the resume paused processes from history menu instead)
Olivier