HI,
Thanx for all the help.
I change the timeo and monitoring the traffics.
Sorry for the late summary.
sc khoo
> Hi Sun Gurus,
> I have two questions I guess you all expert may be able to help.
>
> 1) Recently I have errors on my SUNs whenever some request on NFS
> "NFS read error, RPC time out"
> I was thinking of changing the kernel parameter to extend the time
> so that no time out error. I know this is short term, I am looking
> at the network traffic right now.
>
> 2) How does a SUN print to a printer connected to a terminal server.
> The terminal server is connected to the ethernet back bone.
> Currently a process called "collect" is ruuning to do that. I am
> thinking of printing direct by changing the printcap file.
>
> Any idea?
>
> I will summarize differenly. Thanks a lot.
>
> Swee-Chuan Khoo
> System Administrator
> NS Melaka, Malaysia
1) I never saw that exact error message, so this may be totally off base, but our
NFS timeouts in the past were due to a badly overloaded server. Currently
we're having timeouts which I think are at least partly due to an overloaded
network; we have a switching hub on the way right now which will fix it if
that's the problem.
Steve Ozoa
CAD System Administrator 408-436-4292
ATMEL Corporation fax 408-436-4200
sozoa@atmel.com pager 408-696-0066
2) From: tph@sbctri.sbc.com (Timothy P. Henrion)
We have done with with terminal servers from Xyplex and Xylogics. Both
required a print filter supplied by the vendor. Once you have that
filter, it is fairly simple to update your printcap.
Tim Henrion
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Tim Henrion Southwestern Bell Technology Resources
thenrion@sbctri.sbc.com 550 Maryville Centre Drive
(314)529-7772 St. Louis, MO 63141
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3) P1: Try increasing the timeo and retrans values depending upon the problem
server whereever you specify nfs mounts.
Check the following areas:
-Client Network interfaces
-Network Bandwidth
-Server Network interface
-Server CPU loading
- Server memomry usage
- Server disk bandwidth
-
Try to gather NFS and RPC server/client stats at every hout or so and calculate
teh bad calls timeouts and null calls to identify the bottle necks. Also
if the newtwork utilization is greater than 55% then it's tme to redesign the
network or may be for the time being try adding more memory or faster disks
or upgrading the servers.
Lots of these issues are addressed in Managing NFS and NIS - O'Reilly Associates.
P2: give a IP number to the printer and do printing over the network with
some arbitrary server as print server.
HarisH
Goldman, Sachs NY
(harishm@pcsdnfs1.eq.gs.com)
4) We're printing using tcpxfer (ftp'able from pluto.dss.com, in /pub/tcp_print).
The way it works is you specify either a named pipe or tty in /etc/printcap,
and tcpxfer monitors that device and sends the printout to the terminal server.
I don't know of any more direct way to do it (though if you find one, I
wouldn't mind hearing!)
Steve Ozoa
CAD System Administrator 408-436-4292
ATMEL Corporation fax 408-436-4200
sozoa@atmel.com pager 408-696-0066
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:09:04 CDT