search

ncbi / elastic-blast

ElasticBLAST is a cloud-based tool to perform your BLAST searches faster and make you more effective
https://blast.ncbi.nlm.nih.gov/doc/elastic-blast
Other
46 stars 15 forks source link

No results in output bucket #8

Closed sean-bam closed 2 years ago

sean-bam commented 2 years ago

I am using elastic-blast v. 0.2.3 on a google cloud VM. When I run the tutorial to submit queries from the BDQA WGS project versus refseq_protein, I do not get the expected *.out.gz files. The output bucket contains BDQA/metadata/ folder, so that seems to be populating just fine. If run elastic-blast status --cfg BDQA.ini, I get the following:

Pending 0
Running 0
Succeeded 0
Failed 0

Any ideas why the search doesn't seem to be completing?

christiam commented 2 years ago

Hi @sean-bam , I don't have enough information to diagnose the problem. Do you consistently get the output shown above for the elastic-blast status command? How long has it been since you ran elastic-blast submit ... ? Also, could you please provide us with the information listed in the following page? https://blast.ncbi.nlm.nih.gov/doc/elastic-blast/support.html Thanks, regards,

Christiam

christiam commented 2 years ago

Hi @sean-bam , Did your ElasticBLAST search complete? Thanks, regards,

Christiam

sean-bam commented 2 years ago

Thanks for looking into this. I just tried another run and it finished successfully! So, I will try additional queries today and if the same problem arises I will post the relevant information/logfiles.

However, even though I have the expected outputs, elastic-blast status still reports the job did not finish successfully. Here are the logfiles from this run and responses to your questions:

Do you consistently get the output shown above for the elastic-blast status command?

Yes

How long has it been since you ran elastic-blast submit ...

I execute elastic-blast status as soon as the terminal allows me to do so after I run the submit command (about 3-5 minutes). Here is the relevant info from my environment (I replaced my name and email with something generic)

  1. Configuration file and command line invocation elastic-blast submit --cfg eb.cfg 
    
[cloud-provider]
gcp-project = starlit-myth-339218-scratch
gcp-region = us-east4
gcp-zone = us-east4-b

[cluster] num-nodes = 10 labels = owner=my_username

[blast] program = blastp db = refseq_protein queries = gs://elastic-blast-samples/queries/protein/BDQA01.1.fsa_aa results = gs://starlit-myth-339218-scratch/BDQA options = -task blastp-fast -evalue 0.01 -outfmt "7 std sskingdoms ssciname"



2. system info 
>uname -a 

Linux blast-machine-1 4.19.0-18-cloud-amd64 #1 SMP Debian 4.19.208-1 (2021-09-29) x86_64 GNU/Linux

3. Logfiles
[elastic-blast.log](https://github.com/ncbi/elastic-blast/files/8098108/elastic-blast.log)
[sysconfig.log](https://github.com/ncbi/elastic-blast/files/8098151/sysconfig.log)
[gcloud_info.log](https://github.com/ncbi/elastic-blast/files/8098149/gcloud_info.log)
[env.log](https://github.com/ncbi/elastic-blast/files/8098130/env.log)
christiam commented 2 years ago

hi @sean-bam , Thanks. There is indeed a problem that leads to the misleading output of elastic-blast status when it is immediately run after elastic-blast submit. We are looking into it.

However, even though I have the expected outputs, elastic-blast status still reports the job did not finish successfully.

Do I understand correctly that you can get BLAST output from the results bucket? If so, in this case, what exactly is the output of elastic-blast status that indicates that the job did not finish successfully? Are you seeing this with the example from the quickstart or some other search ?

sean-bam commented 2 years ago

Do I understand correctly that you can get BLAST output from the results bucket?

Yes.

what exactly is the output of elastic-blast status that indicates that the job did not finish successfully?

As you mentioned, if I run elastic-blast status immediately after elastic-blast submit, the output is 

Pending 0
Running 0
Succeeded 0
Failed 0

This led me to believe that the program didn't submit/run/complete. However, now that I know to wait a short period of time to check the status, I can see that elastic-blast is doing its job just fine, and that the problem was with me being impatient :) Sorry for the trouble!

christiam commented 2 years ago

Thanks. There is indeed a problem that leads to the misleading output of elastic-blast status when it is immediately run after elastic-blast submit. We are looking into it.

Hi @sean-bam , I just wanted to let you know that the latest ElasticBLAST release (version 0.2.4) addresses the problem described in my comment above.

Please let us know if you run into any problems and/or have any feedback with this latest release. Thanks!

sbenler commented 2 years ago

Will do, thanks for fixing this small issue!