| Title: | DECnet/OSI for {ULTRIX,OSF/1} |
| Notice: | Indicate version and platform when writing...see #2 for kits |
| Moderator: | BULEAN::CARR |
| Created: | Wed Sep 25 1991 |
| Last Modified: | Thu Jun 05 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 2187 |
| Total number of notes: | 10469 |
We randomly get the following error when establishing approx 30
connections between two systems. The application get errno 60 from
read.
Session control incoming and outgoing timers and osi transport delay
timer have been increased but no effect. There's nothing in syslog
and no swap space warnings.
This problem may coincide with a migration from the LOCAL namespace.
Any ideas what resource is being exhausted?
Thanks,
Stuart.
Environment: DU V3.2D, DNU V3.2A
Event: Local Transport Disconnection
from: Node PENTHOUSE:.osi.ods_lan.fnd4 OSI Transport Local NSAP
%X4900400
8002BA35C0B21 Remote NSAP %X49004008002BA35C0B21
at : 1997-03-06-11:34:25.834+01:00I-----
Reason = Reason not specified
DNA Error = Connection timed out
Event: Local Transport Disconnection
from: Node PENTHOUSE:.osi.ods_lan.fnd4 OSI Transport Local NSAP
%X4900400
8002BA35C0B21 Remote NSAP %X49004008002BA35C0B21
at : 1997-03-06-11:47:32.639+01:00I-----
Reason = Reason not specified
DNA Error = Connection timed out
# ncl sho osi transport local nsap %X49004008002BA35C0B21 Remote NSAP
%X49004008002BA35C0B21 all count
Node 0 OSI Transport Local NSAP %X49004008002BA35C0B21 Remote NSAP
%X49004008002BA35C0B21
AT 1997-03-06-16:13:04.057+01:00I-----
Counters
Failed Checksums = 0
Remote Protocol Errors = 0
Local Protocol Errors = 0
User Octets Received = 10334350
User Octets Sent = 10325018
User PDUs Received = 137094
User PDUs Sent = 137173
Total Octets Received = 11725821
Total Octets Sent = 11735721
PDUs Received = 137094
PDUs Sent = 137173
Duplicate PDUs Received = 21
Retransmitted PDUs = 320
Connects Received = 3678
Connects Sent = 3699
Rejects Received = 0
Rejects Sent = 18
User PDUs Discarded = 0
UD PDUs Received = 0
UD PDUs Sent = 0
Connectionless Bytes Received = 0
Connectionless Bytes Sent = 0
Creation Time = 1997-03-06-11:29:04.432+01:00I-----
on node which issues connects:
Event: Remote Transport Disconnection
from: Node PENTHOUSE:.osi.ods_lan.fndser2 OSI Transport Local NSAP
%X4900
4008002B97008021 Remote NSAP %X49004008002B97008021
at : 1997-03-06-15:46:06.106+01:00I-----
Reason = Reason not specified
Additional Information = '80'H
fnd_server2# ncl sho osi transport Local NSAP %X49004008002B97008021 \
Remote NSAP %X49004008002B97008021 all count
Node 0 OSI Transport Local NSAP %X49004008002B97008021 Remote NSAP
%X49004008002B97008021
AT 1997-03-06-16:46:52.553+01:00I-----
Counters
Failed Checksums = 0
Remote Protocol Errors = 0
Local Protocol Errors = 0
User Octets Received = 94790
User Octets Sent = 94738
User PDUs Received = 1410
User PDUs Sent = 1414
Total Octets Received = 113582
Total Octets Sent = 113653
PDUs Received = 1410
PDUs Sent = 1414
Duplicate PDUs Received = 32
Retransmitted PDUs = 32
Connects Received = 43
Connects Sent = 75
Rejects Received = 1
Rejects Sent = 0
User PDUs Discarded = 0
UD PDUs Received = 0
UD PDUs Sent = 0
Connectionless Bytes Received = 0
Connectionless Bytes Sent = 0
Creation Time = 1997-03-06-14:49:15.409+01:00I-----
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 2139.1 | UPSAR::WALLACE | Digital: A Dilbertian Company | Fri Mar 07 1997 18:08 | 7 | |
Is the server side an appliaction that gets started by the decnet
spawner, or does it create its own listening sockets? Is it the
outgoing timer or incoming timer that is initiating the disconnect?
You can probably tell by bumping the outgoing timer to some huge
number, then seeing if the time for a connect to fail matches
the incoming timer. -- Vince
| |||||
| 2139.2 | disconnect within seconds | CHEFS::DAVIDSONS | Tue Mar 11 1997 18:50 | 107 | |
The server uses dnaOpenIncoming (receive queue increased to 64) and
dnaPollingIncoming.
The disconnects happen within seconds, therefore it's hard to see
whether they're related to session control incoming or outgoing timers.
A 'show osi transport all status' shows max connections increase from 1
to 40 approx while the server application is running.
Note, these problems seem to conincide with a migration from LOCAL to
disributed name space.
Any ideas?
Stuart.
Characteristics of server shown below, client is identical.
(note, NSP not used)
Node 0 OSI Transport
AT 1997-03-11-16:30:56.914+01:00I-----
Characteristics
Maximum Transport Connections = 128
Delay Weight = 255
Delay Factor = 15
Maximum Window = 20
ISO Version = 1
Version = V1.1.0
CONS Classes Supported = { }
CLNS Classes Supported = { 4 }
Support MAP = False
Maximum Remote NSAPs = 160
NSAP Selector = 33
Congestion Avoidance = True
CLTP NSAP Selector = 0
Maximum CLTP Ports = 128
Maximum Listeners = 32
Node 0 Session Control
AT 1997-03-11-16:30:57.991+01:00I-----
Characteristics
Incoming Timer = 360 Seconds
Outgoing Timer = 480 Seconds
Outgoing Proxy = True
Incoming Proxy = True
Address Update Interval = 10 Seconds
Update Retry Interval = 60 Minutes
Version = V3.0.0
Modify ACS = True
Maintain Backward Soft Links = False
Node Synonym Directory = PENTHOUSE:.DNA_NodeSynonym
Soft Link Timer = +30-00:00:00.000I0.000 Days
DECnet Internet Gateway Enabled = False
DECnet Internet Gateway User = ""
Transport Precedence =
(
)
Naming Search Path =
(
[
Directory Service = DECdns ,
Template = "*"
] ,
[
Directory Service = DECdns ,
Template = "PENTHOUSE:.osi.ods_lan.*"
] ,
[
Directory Service = DECdns ,
Template = "PENTHOUSE:.DNA_NodeSynonym.*"
] ,
[
Directory Service = Local ,
Template = "*"
]
)
Backtranslation Search Path =
(
[
Directory Service = DECdns ,
Template = "PENTHOUSE:.DNA_BackTranslation"
] ,
[
Directory Service = Local ,
Template = "*"
]
)
Node 0 Session Control Transport Service osi
AT 1997-03-11-16:31:00.737+01:00I-----
Identifiers
Name = osi
Status
Protocol = '05'H
TSEL = 'DEC0'H
State = On
| |||||
| 2139.3 | UPSAR::WALLACE | Digital: A Dilbertian Company | Wed Mar 12 1997 13:46 | 11 | |
It may be that switching to DECdns has increased the time it
takes to process a connection request, and that the client
side is timing out before it hears anything back from the
server.
What are the osi transport template attributes on the client
side. Initial Retransmit Time and Retransmit Threshold would
be relevant.
Vince
| |||||