When running snippy (toolshed.g2.bx.psu.edu/repos/iuc/snippy/snippy/4.6.0+galaxy0) using a list of dataset pairs, ie paired reads grouped in a paired collection, the tool fails in some of the samples with the following message in the last line: --tmpdir ‘/var/lib/condor/execute/dir_3665157’ is not a directory.
The weird thing is that when I run it for the second time using the same samples in the same paired collection, some samples fail but they are different to the ones that failed in the previous run. That is no say apparently it has nothing to do with the samples as they fail randomly.
Additionally, I tried running the paired fastq files separately one by one, and snippy finishes successfully.
Do you have any clue about what could be going on?
Thanks!!
Sol
Hi @solhaim,
some tools fail sporadically when submitted on a collection. I know it is not really helpful answer. Try the following: re-submit a failed job and at the job setup activate an option that replaces the failed job in collection. I believe it is Attempt to re-use jobs with identical parameters? but I don’t have a failed job on a collection to test it. It works well on jobs with a single output.
Kind regards,
Igor
I have been trying to run snippy on galaxy for a while now (2 months maybe?) and I have the same problem as you. As far as I understood, there was an update on snippy and since then things are weird (A bad cluster node according to the support team). This is the case for both galaxy org and eu.
Going for a different server (Jetstream) didn’t help either. In the end I had to run everything locally.
Hi @solhaim,
found it. It is called ‘Replace elements in collection?’. Set it to Yes during re-run, and if the job completes, the output replaces the failed output in collection.
Hope this helps.
Kind regards,
Igor