Kill nohup bash process

Question

[genomes_and_MGEs]

Hey guys,

I ran this

nohup bash -c 'for next in $(cat all_bacteria_links.txt); do wget -P all_bacteria "$next"/*genomic.fna.gz; done',

and now I want to kill the process. I ran the command in ssh server, so whenever I try to kill PID or kill -9 PID, the server connection closes. Can you please let me know how I can kill this process? Thanks!

[Santosh Anand]

Not a bioinfo Q. Still, this might help you https://stackoverflow.com/questions/17385794/how-to-get-the-process-id-to-kill-a-nohup-process

[Devon Ryan]

You're killing the wrong bash process.

[WouterDeCoster]

Do ps -ef | grep all_bacteria_links and kill the right process.

[Ram]

Pro tip: use a decoy character class with the grep following the ps to avoid listing the grep as one of the candidates.

ps -ef | grep "[a]ll_bacteria_links"

The pattern [a]ll doesn't match the string "[a]ll", so the grep process won't be picked.

[Santosh Anand]

OP has a very convoluted way to run the command by running a bash shell with nohup, so the process to kill is bash.

[WouterDeCoster]

I believe there is a specific corner in hell for people who do this, no?

[genomes_and_MGEs]

It doesn't work, because the PID is always changing on the server... So, I found a solution:

pkill -STOP -P the_ppid

Thanks anyway guys!

Answer 1

[shenwei356]

Hi, I'd recommend using screen instead of nohup, screen is safer.

And, you better utilize some batch commands like parallel to accelerate thousands of jobs, here is wget.

Using wget -c to resume unfinished download is also recommended.

Here's my method in this case from Manipulation on NCBI refseq bacterial assembly summary

$ cat mt.tsv | csvtk cut -t -f ftp_path | sed 1d \
    | rush -v prefix='{}/{%}' \
        '   wget -c {prefix}_genomic.fna.gz; \
            wget -c {prefix}_genomic.gbff.gz; \
            wget -c {prefix}_genomic.gff.gz; \
            wget -c {prefix}_cds_from_genomic.fna.gz \
            wget -c {prefix}_protein.faa.gz; \
        ' \
        -j 10 -c -C download.rush

If the process died for some reasons like PC reboot or network interrupt, just re-run the command, which will ignore finished jobs recorded in file download.rush.

[Santosh Anand]

THIS is the right solution!