From: Res <res(a)ausics.net>
> I have a problem with a CentOS 4 and now updated to 4.1 server,
> every so often the other servers try to contact it and this not there a
> retry and it is, like it goes to sleep, I have no problem with the
> fedora1, RH9 and slackware machines they just never do this.
> its most annoying..
> they all use IP's not names. all access lists are identical, all hosts
> files are identical, all exports and fstabs are same format.
> in fact one of the machines on another network altogether reponds
> instantly. if I reboot *shudder* the CentOS box it fixes it, temporarily,
> but thats a joke, the hardware is fine, it used to run slackware
> smoothly, I hope I dont have to go back to it :)
What are your export options?
What are your mount options?
Have you considered using the automounter (or are you already)?
From: Johnny Hughes <mailing-lists(a)hughesjr.com>
> I haven't seen this issue ... but it might actually be sleeping.
> CentOS-4 uses a 2.6.9-x kernel that is much better at doing what your
> BIOS says for power saving mode than 2.4 kernels were. I would first
> check the sleep settings in the BIOS of the problem machine.
> If that doesn't work, you try turning off the apmd and acpid services.
> Also, to really search for fixes ... the exact hardware that is having
> problems might help.
I typically find NFS bliss if I stick with a client and server in the same
series as follows:
- 2.2.18+ (or early 2.2.x with Trond+Higgen, thank you SGI!)
- 2.4.9 or earlier (VM #1)
- 2.4.10-2.2.18 (VM #2)
- 2.4.21 or later (much better NFS client IMHO)
- 2.6.x or later (thank you Sun!)
I always have issues if I start using one series of Linux NFS clients
with another series of Linux NFS servers, etc...
In your case, your server is 2.6.x, and your clients are 2.4.18, maybe
2.4.21+. Those 2.4.19 and 2.4.20 releases were "less than fun"
when it came to the Linux NFS client, especially when not connecting
to a NFS server of the same series.
--
Bryan J. Smith mailto:b.j.smith@ieee.org