[Labs-l] NFS issues, please deploy fire

[Petr Bena]

Actually, there are issues with gluster which isn't much better, so
any alternative to gluster or nfs would be nice

On Thu, Sep 13, 2012 at 8:33 PM, Damian Zaremba
<damian at damianzaremba.co.uk> wrote:
> As seen on a few instances today;
> [  825.247230] svc: failed to register lockdv1 RPC service (errno 97).
> [463494.148261] svc: failed to register lockdv1 RPC service (errno 97).
> [1442873.402458] svc: failed to register lockdv1 RPC service (errno 97).
>
> Logins also seem to be rather laggy, ganglia shows spiky cpu on the nfs
> instance.
>
> Any chance of moving home dirs over to the same cluster as project storage
> any time soon?
>
> NFS isn't really scalable in the current state and is rather laggy randomly,
> also if that instance where to die everyone would be screwed.
>
> Considering that opening up registration also potentially increases active
> users, it might be an idea to solve the issue before it seriously becomes
> one and we have a whole gluster like downtime.
>
> TL;DR NFS should DIAF and we should use gluster/ceph for user data.
>
> Damian
> --
> Damian Zaremba
> http://about.me/damianzaremba
>
>
> _______________________________________________
> Labs-l mailing list
> Labs-l at lists.wikimedia.org
> https://lists.wikimedia.org/mailman/listinfo/labs-l