I'm not sure if it's worth pursuing under the assumption that storage
stability will happen in the immediate future, but...

What if, instead of mounting the keys with NFS; we stored the keys locally.
IE: Distribute them with rsync or what we're considering for deployment on
the cluster?

~Matt Walker
Wikimedia Foundation
Fundraising Technology Team


On Wed, Mar 6, 2013 at 12:00 PM, Petr Bena <[email protected]> wrote:

> Okay and that's my point. Problem is only on 1 instance or multiple
> instances but not on all of them. So being able to get into bastion
> alternative way doesn't mean you won't get elsewhere
>
> On Wed, Mar 6, 2013 at 8:58 PM, Ryan Lane <[email protected]> wrote:
> > On Wed, Mar 6, 2013 at 11:50 AM, Petr Bena <[email protected]> wrote:
> >>
> >> But I already replied to that once.
> >>
> >> If this problem was affecting all instances - why bastion2 was working
> >> hm? it's a local problem - I don't believe we wouldn't get into other
> >> instances, I was able to ssh to all of them from bastion when my
> >> session was still open while others just couldn't get into bastion
> >>
> >
> > SSH keys are NFS. All NFS access goes to a specific host. Glusterfs has
> an
> > issue where a single client can have issues connecting while others do
> not,
> > but it doesn't really work that way with NFS.
> >
> > - Ryan
> >
> > _______________________________________________
> > Labs-l mailing list
> > [email protected]
> > https://lists.wikimedia.org/mailman/listinfo/labs-l
> >
>
> _______________________________________________
> Labs-l mailing list
> [email protected]
> https://lists.wikimedia.org/mailman/listinfo/labs-l
>
_______________________________________________
Labs-l mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/labs-l

Reply via email to