Bug#875925: hdf5: interaction with nfs changed between jessie and stretch

[Gilles Filippini]

Hi Frédéric,

picca at synchrotron-soleil.fr a écrit le 16/09/2017 à 07:25 :
> Source: hdf5
> Version: 1.10.0-patch1+docs-3
> Severity: important
> 
> Dear Maintainer,
> 
> after upgrading my system from jessie to stretch, I could not read anymore my hdf5 files located on
> our nfs filesystem.
> 
> I have an error like this 
> 
> Can not flock ...
> 
> looking at the nfs documentation, I could find flock in the local_lock options which is set to
> "none" on my system.
> 
> Since I red that some of these options can create conflict between lock mechanism.
> And because I am not the only one accessing these files (other system are CentOS 5, 6).
> I would like your opinion on the best way to solve this issue.

Well, honestly I'm not at ease at all with file locking over NFS. Would
you mind forwarding this question to upstream [1]?

[1] help at hdfgroup.org

As I understand it, you should experience the very same problem using
flock(1) shell command on these files. Is that correct?

Thanks,

_g.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 508 bytes
Desc: OpenPGP digital signature
URL: <http://lists.alioth.debian.org/pipermail/pkg-grass-devel/attachments/20170916/0349d85d/attachment.sig>