(I've tried to put relevant keywords in the subject header...
And I apologize for the length of the message, but sometimes people leave out the
crucial bits, and I'm trying to avoid this.)
We have a bunch of rackmount AlphaServer 1000a 5/300's, and a 2100a 5/300, pedestal model.
All machines have worked fine with their network cards running at 10Mbits; the problems have come
about at 100Mbits. There are 3 separate problems that I've encountered, which stump me. The first
seems to be a hardware level issue; the 2nd and 3rd may be more software related, perhaps.
Problem 1. One of the 1000a's, with DE500-XA network cards in slots 12 and 13 (primary PCI bus), ran
for weeks fine on a 10Mbit segment. Moving to the computer room, and giving it a 100Mbit switched
port on the hub, setting ewa_mode to fast, it will NOT get a link light on the card in slot 12. We've done the
obvious things: switched with known good patches etc. The card tests OK, (see question 1 below),
and the other card gets link also. Identically configured machines have no problem. I'm a little reluctant
to believe the card is bad, but maybe so. See after the .signature for console output.
Question 1.1: Are there more tests that I can run from the console level of the cpu?
(Digital's install booklet talks about using EzWorks, some PC based application,
to test and configure the network cards. Does anyone know how to run such tests on a
real machine?? )
Question 1.2: Is there a general problem with the DE500-XA's? It would appear that all the
DE500-AA improves is with the autosensing mode (useful) and the autoboot ROM
support (useless), is there anything else?
Question 1.3: We're initially doing half-duplex at 100Mbit, has anyone experience with full-duplex
mode?
Problem 2. An identically configured machine to that in Problem 1. is also connected to
2 100Mbit networks. It gets link lights on power up; the network tests turn the link off
and on briefly. The operating system is Digital Unix 4.0b, and on boot up the link light on tu0
goes OFF (see the uerf output below, note the "no link" on tu0 message). There is a little
attempt at generating net activity with ntpdate, then ypbind runs, and by the time an NFS
mount is attempted over this interface, the link comes on, and the rest is all normal from
there on. Note that tu1's link light never died.
So this machine is functional but I'm not happy about this.
I've gotten both the huge Feb and April 4.0b patches, neither discuss this issue.
Question 2.1: Is this a real problem, i.e., is it a bug in DU4.0b, and will it cause
any production problems? (That is, will the machine arbitrarily change link states...)
Problem 3. The 2100a we have functioned excellently for several months running DU3.2g, with a
single DE435 10Mbit ethernet card. I got the upgrade fever; and the machine was
reconfigured to run 4.0b, with a firmware upgrade to 4.7 done, new SCSI controller
cards, and 2 DE500-XA's installed. See the console output for Problem 3 section
for firmware details, and where the cards are installed. There is no link light problem
on powerup, both cards see their 100Mbit networks. The same problem as in Problem 2
happens with the primary interface tu0 (link off during boot; comes back on after
some attempts to generate network activity during boot). However this is only the
beginning.
The real problem is that when one logs into the machine over tu0, and does anything
that generates network activity (e.g., cd /usr; ls -lR with output to an rsh session), after
about thirty lines of output or less, the machine appears to freeze, losing all network
activity for about a minute or more, and this message appears on the console:
tu0: transmit FIFO underflow: threshold raised to: 256 bytes
After this life resumes; but unpredictively later, one gets a similar message, but
with 256 replaced by 512. It wouldn't be so bad if this only happened once, but
I simply cannot afford to have a production machine freeze for minutes at a time.
Question 3.1: Does anyone know what's happening and what to do?
I've gotten both the huge Feb and April 4.0b patches, neither discuss this issue.
(And I do know how to set the Fast mode console variables...)
(I'm also calling Digital CSC about these problems; the usual test applies: will the mailing list
come through before Digital or not?)
More questions: 4) our supplier is reporting multi-week delays in getting hold of any kind of
DE500 cards;
does anyone know of a quick source, or alternate 100Mb ethernet cards for
AlphaServers which work under Digital Unix 4.0b?
5) Are these problems because we have DE500-XA cards instead of DE500-AA's?
6) Is there firmware to take care of this?
Thanks in advance for any ideas or sharing of related experiences.
Oisin McGuinness
Sumitomo Bank Capital Markets
277 Park Avenue
New York NY 10172
USA
(212)-224-4913, email: oisin_at_sbcm.com
--------------- Console information for problem 1 ----------------------
Digital Equipment Corporation
AlphaServer 1000A 5/400
Firmware
SRM Console: V4.7-169
ARC Console: 5.23
PALcode: VMS PALcode V1.19-3, OSF PALcode V1.21-5
Serial Rom: V1.0
Processor
DECchip (tm) 21164A-2 400MHz
Memory
544 Meg of System Memory
Bank 0 = 32 Mbytes(8 MB Per SIMM) Starting at 0x20000000
Bank 1 = 256 Mbytes(64 MB Per SIMM) Starting at 0x00000000
Bank 2 = 256 Mbytes(64 MB Per SIMM) Starting at 0x10000000
Bank 3 = No Memory Detected
Slot Option Hose 0, Bus 0, PCI
7 Intel 82375EB Bridge to Bus 1, EISA
8 DECchip 21050-AA Bridge to Bus 2, PCI
11 S3 Trio64/Trio32
12 DECchip 21140-AA ewa0.0.0.12.0 00-00-F8-1E-93-4D
13 DECchip 21140-AA ewb0.0.0.13.0 00-00-F8-1E-87-42
Slot Option Hose 0, Bus 1, EISA
Slot Option Hose 0, Bus 2, PCI
0 QLogic ISP1020 pka0.7.0.2000.0 SCSI Bus ID 7
dka0.0.0.2000.0 RZ26N
dka100.1.0.2000.0 SEAGATE ST31051W
dka500.5.0.2000.0 RRD45
1 DECchip 21050-AA Bridge to Bus 3, PCI
2 DECchip 21050-AA Bridge to Bus 4, PCI
Slot Option Hose 0, Bus 3, PCI
0 QLogic ISP1020 pkb0.7.0.3000.0 SCSI Bus ID 7
Slot Option Hose 0, Bus 4, PCI
0 QLogic ISP1020 pkc0.7.0.4000.0 SCSI Bus ID 7
>>>show ewa*mode
ewa0_mode Fast
>>>show ewb*mode
ewb0_mode Fast
>>>
--------------------- UERF report for Problem 2 --------------------------------
MESSAGE Alpha boot: available memory from
_0x1240000 to 0x27fee000
Digital UNIX V4.0B (Rev. 564); Wed
_Apr 2 08:42:17 EST 1997
physical memory = 640.00 megabytes.
available memory = 621.71 megabytes.
using 816 buffers containing 6.37
_megabytes of memory
Firmware revision: 4.7
PALcode: OSF version 1.21
AlphaServer 1000A 5/400
Mem_size = 669138944, consmem = 238
pci0 at nexus
eisa0 at pci0
ace0 at eisa0
ace1 at eisa0
lp0 at eisa0
fdi0 at eisa0
fd0 at fdi0 unit 0
pci2000 at pci0 slot 8
isp0 at pci2000 slot 0
isp0: QLOGIC ISP1020A
isp0: Firmware revision 2.10 (loaded
_by console)
scsi0 at isp0 slot 0
rz0 at scsi0 target 0 lun 0 (LID=0)
_(DEC RZ26N (C) DEC 0616)
_(Wide16)
rz1 at scsi0 target 1 lun 0 (LID=1)
_(SEAGATE ST31051W 0526)
_(Wide16)
rz2 at scsi0 target 2 lun 0 (LID=2)
_(SEAGATE ST31051W 0526)
_(Wide16)
rz5 at scsi0 target 5 lun 0 (LID=3)
_(DEC RRD45 (C) DEC 0436)
pci3000 at pci2000 slot 1
isp1 at pci3000 slot 0
isp1: QLOGIC ISP1020A
isp1: Firmware revision 2.10 (loaded
_by console)
scsi1 at isp1 slot 0
rz8 at scsi1 target 0 lun 0 (LID=4)
_(SEAGATE ST32171W 0280)
_(Wide16)
rz9 at scsi1 target 1 lun 0 (LID=5)
_(SEAGATE ST32171W 0280)
_(Wide16)
pci4000 at pci2000 slot 2
isp2 at pci4000 slot 0
isp2: QLOGIC ISP1020A
isp2: Firmware revision 2.10 (loaded
_by console)
scsi2 at isp2 slot 0
rz16 at scsi2 target 0 lun 0 (LID=6)
_(SEAGATE ST32171W 0280)
_(Wide16)
rz17 at scsi2 target 1 lun 0 (LID=7)
_(SEAGATE ST32171W 0280)
_(Wide16)
trio0 at pci0 slot 11
trio0: S3 Trio64 (SVGA) - Plug N' Play
_- 2.0 Mb
tu0: DECchip 21140-AA: Revision: 2.0
tu0: auto negotiation capable device:
_National DP83840
tu0 at pci0 slot 12
tu0: DEC Fast Ethernet Interface,
_hardware address: 00-00-F8-03-6E-BE
tu0: auto negotiation off: selecting
_100BaseTX (UTP) port: half duplex:
_no link ************************************
tu1: DECchip 21140-AA: Revision: 1.2
tu1 at pci0 slot 13
tu1: DEC Fast Ethernet Interface,
_hardware address: 00-00-F8-1E-82-9C
tu1: console mode: selecting 100BaseTX
_(UTP) port: half duplex
gpc0 at eisa0
No FRU table entry found in
_console-rpb
--Not creating a FRU table errorlog
_packet
lvm0: configured.
lvm1: configured.
kernel console: ace0
dli: configured
-------------------------- UERF for Problem 3 -------------------------
MESSAGE Alpha boot: available memory from
_0xedc000 to 0x1ffee000
Digital UNIX V4.0B (Rev. 564); Fri
_May 16 15:28:23 EDT 1997
physical memory = 512.00 megabytes.
available memory = 497.07 megabytes.
using 652 buffers containing 5.09
_megabytes of memory
Master cpu at slot 0.
Firmware revision: 4.7
PALcode: OSF version 1.21
ibus0 at nexus
AlphaServer 2100A 5/300
cpu 0 EV-5 4mb b-cache
cpu 1 EV-5 4mb b-cache
gpc0 at ibus0
pci0 at ibus0 slot 0
eisa0 at pci0
ace0 at eisa0
ace1 at eisa0
lp0 at eisa0
fdi0 at eisa0
fd0 at fdi0 unit 0
pci2000 at pci0 slot 3
psiop0 at pci2000 slot 1
Loading SIOP: script 1007e00, reg
_81900000, data 1013d40
scsi0 at psiop0 slot 0
rz0 at scsi0 target 0 lun 0 (LID=0)
_(DEC RZ28D (C) DEC 0008)
_(Wide16)
rz1 at scsi0 target 1 lun 0 (LID=1)
_(DEC RZ28D (C) DEC 0010)
_(Wide16)
rz2 at scsi0 target 2 lun 0 (LID=2)
_(DEC RZ28D (C) DEC 0010)
_(Wide16)
rz6 at scsi0 target 6 lun 0 (LID=3)
_(DEC RRD45 (C) DEC 0436)
tu0: DECchip 21140-AA: Revision: 1.2
tu0 at pci2000 slot 6
tu0: DEC Fast Ethernet Interface,
_hardware address: 00-00-F8-03-C9-D7
tu0: console mode: selecting 100BaseTX
_(UTP) port: half duplex: no link *******************************************
psiop1 at pci2000 slot 7
Loading SIOP: script 1019e00, reg
_81900100, data 40600140
scsi3 at psiop1 slot 0
tz30 at scsi3 target 6 lun 0 (LID=4)
_(DEC DLT2000 840F)
pci3000 at pci2000 slot 8
isp0 at pci3000 slot 0
isp0: QLOGIC ISP1020A
isp0: Firmware revision 2.10 (loaded
_by console)
scsi2 at isp0 slot 0
rz16 at scsi2 target 0 lun 0 (LID=5)
_(SEAGATE ST32171W 0280)
_(Wide16)
rz17 at scsi2 target 1 lun 0 (LID=6)
_(SEAGATE ST34371W 0280)
_(Wide16)
trio0 at pci0 slot 6
trio0: S3 Trio64 (SVGA) - Plug N' Play
_- 1.0 Mb
tu1: DECchip 21140-AA: Revision: 1.2
tu1 at pci0 slot 7
tu1: DEC Fast Ethernet Interface,
_hardware address: 00-00-F8-03-D3-C6
tu1: console mode: selecting 100BaseTX
_(UTP) port: half duplex
pci4000 at pci0 slot 8
isp1 at pci4000 slot 0
isp1: QLOGIC ISP1020A
isp1: Firmware revision 2.10 (loaded
_by console)
scsi1 at isp1 slot 0
rz8 at scsi1 target 0 lun 0 (LID=7)
_(SEAGATE ST32171W 0280)
_(Wide16)
rz9 at scsi1 target 1 lun 0 (LID=8)
_(SEAGATE ST34371W 0280)
_(Wide16)
kernel console: ace0
-------------------- CONSOLE INFO for Problem 3 ----------------
P00>>>show conf
Digital Equipment Corporation
AlphaServer 2100A 5/300
SRM Console V4.7-157 VMS PALcode V1.19-2, OSF PALcode V1.21-1
Component Status Module ID
CPU 0 P B2040-BA DECchip (tm) 21164-5
CPU 1 P B2040-BA DECchip (tm) 21164-5
Memory 0 P B2022-FA 512 MB
I/O B2222-AA
dva0.0.0.1000.0 RX26/RX23
Slot Option Hose 0, Bus 0, PCI
2 Intel 82375EB Bridge to Bus 1, EISA
3 DECchip 21050-AA Bridge to Bus 2, PCI
6 S3 Trio64/Trio32
7 DECchip 21140-AA ewb0.0.0.7.0 00-00-F8-03-D3-C6 *********************
8 DECchip 21050-AA Bridge to Bus 4, PCI
Slot Option Hose 0, Bus 1, EISA
Slot Option Hose 0, Bus 2, PCI
1 NCR 53C810 pka0.7.0.2001.0 SCSI Bus ID 7
dka0.0.0.2001.0 RZ28D
dka100.1.0.2001.0 RZ28D
dka200.2.0.2001.0 RZ28D
dka600.6.0.2001.0 RRD45
6 DECchip 21140-AA ewa0.0.0.2006.0 00-00-F8-03-C9-D7 *****************
7 NCR 53C810 pkb0.7.0.2007.0 SCSI Bus ID 7
mkb600.6.0.2007.0 DLT2000
8 DECchip 21050-AA Bridge to Bus 3, PCI
Slot Option Hose 0, Bus 3, PCI
0 QLogic ISP1020 pkc0.7.0.3000.0 SCSI Bus ID 7
dkc0.0.0.3000.0 SEAGATE ST32171W
dkc100.1.0.3000.0 SEAGATE ST34371W
Slot Option Hose 0, Bus 4, PCI
0 QLogic ISP1020 pkd0.7.0.4000.0 SCSI Bus ID 7
dkd0.0.0.4000.0 SEAGATE ST32171W
dkd100.1.0.4000.0 SEAGATE ST34371W
P00>>>
------------------ That's all folks, thanks for reading this far --------------------
Received on Tue May 20 1997 - 16:40:28 NZST