Closed innovate-invent closed 4 years ago
What is used to generate the parsnp binary in the release? It seems as if it is not unpacking the bin folder from itself.
Either the program deletes /tmp/_MEItORVk1/bin
on failure or the folder is never created.
Causing the symlinks to be broken:
ls -l /tmp/_MEIhjQF5K/
total 0
lrwxrwxrwx 1 root root 34 Feb 4 18:19 ft -> /tmp/_MEIhjQF5K/bin/fasttree_linux
lrwxrwxrwx 1 root root 33 Feb 4 18:19 harvest -> /tmp/_MEIhjQF5K/bin/harvest_linux
lrwxrwxrwx 1 root root 29 Feb 4 18:19 nucmer -> /tmp/_MEIhjQF5K/MUMmer/nucmer
lrwxrwxrwx 1 root root 26 Feb 4 18:19 parsnp -> /tmp/_MEIhjQF5K/bin/parsnp
lrwxrwxrwx 1 root root 33 Feb 4 18:19 phiprofile -> /tmp/_MEIhjQF5K/bin/Profile_linux
lrwxrwxrwx 1 root root 34 Feb 4 18:19 show-coords -> /tmp/_MEIhjQF5K/MUMmer/show-coords
It appears these issues do not occur on a mac. In linux building from source and running Parsnp.py with python2 seemed to get a bit further but returns the same error.
Manually running /tmp/_MEItORVk1/parsnp /home/nolan/Downloads//parsnpAligner.ini
fails with:
*** ERROR *** DistKmer4_6 requires nucleo alphabet
*** ERROR *** MSA::GetLetter(3/4, 1/11)='T'/4294967295
*** ERROR *** MSA::GetLetter(0/5, 4/12)='G'/4294967295
Running it a few more times actually generates a different set of errors. Mostly DistKmer4_6 errors but it actually seg faulted once.
It seems the problem is specific to my computer:
uname -a
Linux ASUS-N580GD 4.20.0-arch1-1-ARCH #1 SMP PREEMPT Mon Dec 24 03:00:40 UTC 2018 x86_64 GNU/Linux
Running ParSNP inside a Debian docker instance fails just the same, but if I run it in a KVM ubuntu image it doesn't fail. I can only speculate that KVM isn't forwarding some CPU feature that libMUSCLE fails to use.
same error, same error. all arch, arch-related distro and docker in arch failed both package/conda and compiling installation . and runing parsnp was fine in ubuntu.
Ahh. I've also seen this issue. Has something to do with OpenMP. We have resolved this in our current builds, but depending on your build system you may still get this error. Please let me know.
Installing ParSNP from conda doesn't seem to work. I get the following error running on some basic Pseudomonas genomes taken from NCBI.
This seems to be the same problem as #35 but I am creating a new issue as that one has not been addressed in several years.