Thanks to
Olle Eriksson
Iwao MAKINO
Joe Fletcher
Dr. Tom Blinn
Answers/Suggestions:
Cpu errors are probably related to the memory DIMMS.
Full-duplex on a 21140 should be OK.
Install latest patch kit.
- I swapped in a new 256MB kit and all is well for now. We'll see how it
goes. I have two spare kits & it still would not work with two, identical
kits installed; same panic message. These are DEC kits.
- Running at 100Mb full, collision rate is in the dirt, response fabulous.
- I'll have to do the patch kit at a later date & test it with the 2nd
memory kit.
Slipped my mind to mention this before, I added the following to
/etc/sysconfigtab last night per
http://www.unix.digital.com/internet/tuning.htm#recommend
vm:
vm-maxvas = 4398046511104
vm-mapentries = 20000
vm-ubcseqpercent = 20
vm-mapentries = 16384
vm-vpagemax = 536870912
proc:
max-per-proc-address-space = 4398046511104
max-per-proc-data-size = 4398046511104
per-proc-address-space = 4398046511104
per-proc-data-size = 4398046511104
maxusers = 2048
max-proc-per-user = 512
max-threads-per-user = 4096
socket:
somaxconn = 32767
sominconn = 32767
sbcompress_threshold = 600
ipport_userreserved = 65000
inet:
tcbhashsize = 16384
pmtu_enabled = 0
Original message . . . .
>
> Last night I made a couple changes to our web/mail/nis/router server, an
>
> AlphaStation 500/400 w/ 4.0E patch 1, but I now have some questions & a
> problem.
>
> Swapped 128 MB memory kit for a 256 MB kit. The kits wouldn't co-exist,
> system
> panic'd w/only 33MB physical memory, or similar verbage.
> ---Shouldn't they co-exist? Maybe the 2nd set of SIMM slots are
> defective?
>
> Has 2 PCI (21140 chipset) ethernets installed & routed to those subnets,
> but I
> wanted to remove 1 & switch to 100Mb half duplex so I: set ewb0_mode
> Fast,
> changed NETDEV_0="tu1", NETDEV_1="tu2", tu0 is not assigned an IP in
> /etc/rc.config. This is happy so far, plugged into a BayNetworks Switch
> 28115.
> ---Is it safe to go to 100Mb full w/ the 21140? I have a 21143 card I
> could put
> in.
>
> I was done at 7:30PM last night but early morning, the CPU started
> logging
> errors.
>
> Mar 24 03:39:16 hytek vmunix: WARNING: too many Processor corrected
> errors
> detected on cpu 0. Reporting suspended.
>
> uerf shows it started at around 2AM. It's logged more than 100 entries.
> ---How can I determine the source of the CPU problem? Is there a listing
> somewhere
> for the errors codes?
>
> ********************************* ENTRY 1.
> *********************************
>
> ----- EVENT INFORMATION -----
>
> EVENT CLASS ERROR EVENT
> OS EVENT TYPE 100. CPU EXCEPTION
> SEQUENCE NUMBER 154.
> OPERATING SYSTEM DEC OSF/1
> OCCURRED/LOGGED ON Fri Mar 24 08:11:54 2000
> OCCURRED ON SYSTEM hytek
> SYSTEM ID x0007000F
> SYSTYPE x00000000
>
> ----- UNIT INFORMATION -----
>
> UNIT CLASS CPU
>
> ********************************* ENTRY 2.
> *********************************
>
> ----- EVENT INFORMATION -----
>
> EVENT CLASS ERROR EVENT
> OS EVENT TYPE 100. CPU EXCEPTION
> SEQUENCE NUMBER 153.
> OPERATING SYSTEM DEC OSF/1
> OCCURRED/LOGGED ON Fri Mar 24 08:11:54 2000
> OCCURRED ON SYSTEM hytek
> SYSTEM ID x0007000F
> SYSTYPE x00000000
>
> ----- UNIT INFORMATION -----
>
> UNIT CLASS CPU
>
--
Toby Bluhm
tkb_at_mr.marconimed.com
440-483-5323
Received on Fri Mar 24 2000 - 19:23:08 NZST