We had an issue recently where our NAS needed to be powered down for some building maintenance, a couple client machines had some problems.
Long story short, a couple of Hadoop servers that use a backend, point-to-point interface to the NAS didn't come up, because of a mistake in NIC settings on the NAS. Once those were resolved, things slowing began to start coming back; ultimately, I rebooted them.
We are using RHEL and CentOS 5.x and 6.x in our environment. The NAS is Solaris 10 (thumper class).
I use these options for the mount points "tcp,hard,intr,rw,bg" where I expected that if for some reason there was an issue (according to the docs) the mount process would be backgrounded and the server would just come up. This doesn't appear to be the case and I'm hearing that the Linux NFS client doesn't handle this so gracefully.
I wonder if there is a way to mitigate this in the future. Any suggestions would be appreciated.
Thank you.
Long story short, a couple of Hadoop servers that use a backend, point-to-point interface to the NAS didn't come up, because of a mistake in NIC settings on the NAS. Once those were resolved, things slowing began to start coming back; ultimately, I rebooted them.
We are using RHEL and CentOS 5.x and 6.x in our environment. The NAS is Solaris 10 (thumper class).
I use these options for the mount points "tcp,hard,intr,rw,bg" where I expected that if for some reason there was an issue (according to the docs) the mount process would be backgrounded and the server would just come up. This doesn't appear to be the case and I'm hearing that the Linux NFS client doesn't handle this so gracefully.
I wonder if there is a way to mitigate this in the future. Any suggestions would be appreciated.
Thank you.
