Telnet session disconnects abruptly

I have inherited and SCO OpenServer Release 6 server. The clients connect using telnet to get to a proprietary database application for Service tickets. The issue I am currently having is that the connection just stops abruptly and you can see "telnet session terminated" on the terminal emulation software. The terminal emulation software (Attachmate Relections) has a trace function and I can see an received FD_Close from the server. I have tried a wireshark trace as well and can see several TELNET retransmissions before the disconnect. This has happened with two different terminal emulation programs and on different computers on the network. I have checked /usr/adm/messages and there is nothing relating to the telnet sessions, and I have checked /usr/adm/syslog and I only get the following messages

service bootpd[2175]: IP address not found: xxx.xxx.xxx.xxx 

Where xxx.xxx.xxx.xxx shows the IP address of the computer connecting.

Is there anyway to find out why SCO is disconnecting the sessions. There is no rhyme or reason I can see. It can occur after 1 minute or after 1 hour. If I have two terminal emulator session running on the computer at the same time one session is disconnected and the other keeps running fine for the rest of the day.

I've tried the switch but don't see any errors?

Would anyone have any ideas on why SCO appears to be dropping the connection and where I can look?
Thank you

There are people on this forum who know SCO much better than me as I've not used SCO for years.

However, I do remember that the kernel has "Streams" buffers configured for various network packet sizes and if these buffers are max'd out all hell breaks loose on the network.

I remember you can look at these using the command:

# crash

and then subcommand:

 
 strstat
 

meaning streams statistics.

If these buffers need increasing it involves running ./configure and then relinking the kernel, followed by a reboot to bring that kernel in.

As I say, you will probably get much better answers from others but I hope that helps.

Do you have "keepalive" set on in the terminal emulator?
Does the same thing happen if you switch to ssh?

Thanks hicksd8, I have checked the streams and it appears to be ok

# crash
dumpfile = /dev/mem, namelist = /stand/unix, outfile = stdout
Engine: 0 of 4  Procslot: 80  Lwpslot: 0  crash
> strstat
ITEM                  INUSE      TOTAL       FAIL
streams                 215     155432          0
queues                 1066     615058          0
message blocks            0     739991          0
message triplets        390    9538784          0
link blocks              26         43          0
stream events             0          0          0

I don't see any fails.

Thank you JGT,
I do not see a keep alive setting in terminal emulation software, but since the disconnect can happen at any time would this be the keep alive? I have seen the same user with an issue leave for lunch for 30 minutes and the session stay up. I have also seen the same user loose connection in the middle of typing on the computer.
I have looked into changing ssh but the proprietary database application for Service tickets does not open when connecting to the ssh session it drops the user into the command prompt and not the application, so I have not been able to connect using ssh for the application.

Several thoughts.
Can you download and install Welcome to April System Design
Pick a version that suits. There is a keep alive setting on the "Config->general settings" window.
How is the user application program started? By setting the shell in etc/passwd, or by 'exec pgm' as the last line of the .profile?
If by the last line of the .profile can you change the .profile so that if the application program dies, the user drops out to a $ prompt instead of being logged off?

Thanks for the reply jgt,
I have downloaded ICE TCP Plus from James River to try a different Terminal emulator. I installed it and will run it on the user's computer having the most trouble.
At this point I am not sure how the program gets started, my apologies but SCO is very new to me I am reading and learning at the same time. I will look this up and get back to you. The problem is when the user drops they lose what they were typing at the time and for some users this could be several lines of a document, so if they are logged off or dropped to the prompt I think they will still lose the information, which is the big problem.
Would you have a link that I can read regarding the .profile and /etc/passwd files and how they work so I can try to figure out how they access the application?

Password file passwd(F)

.profile profile(F)

The entire document library is installed on your local system by default. You can access it by pointing your browser to the SCO host.
Are there any routers/firewalls between the users and the SCO system?

jgt,
Since running the other Terminal emulation software the link has stayed up all day. I have not changed anything in the default configuration on the ICE terminal package except for the terminal type for SCO ANSI and the screen lines needed to be 25 instead of the default 24. Nothing has been changed on the SCO Server right now. I will continue to let it run for a few days and see what happens.
I appreciate all of the help and guidance that I have received

I have tried to install the other Terminal Emulator program and it worked well for 1 day, and as of today the Server has disconnected the Terminal Emulator ICE TCP Plus about 5 times.
In answer to jgt there is no firewall or router in between the users and the SCO system.
Would anyone have any ideas why the telnet session just terminates. It appears to be coming from the server. Is there any debug or monitoring I can turn on to get the information as to why the connection is being terminated?
I appreciate any help.
Thank you

Do you have devices with static ip addresses within the DHCP range?

Changing telnet client will almost certainly not fix this problem for you. That s a red herring.

My question for you - is this actually a new problem for your users or is this an old recurring problem which your users have lived with for a long time? If it is a new problem, what has changed?

jgt,
I have checked an no there are no static addresses in the DHCP range. I have done a span port to wireshark trace the data in and out of the server and do not see any conflicts, nor am I seeing any of the windows clients responding with duplicate IP address errors.

fpmurphy,
Thank you I have been trying to go over what could have changed in the last month when this appears to have started, as it is new. The only difference at the moment is that the user computers used to be static IP addresses and I have replaced a number of them as they were old. When I found out they were static I changed them to DHCP (I changed them because I found that the wireless was issuing DHCP addresses in the static range and so I changed all devices to static except for the servers and routers). Could changing the user computers to DHCP effect the server telnet connections? Could SCO block connections from specific addresses? If it were to block connections from certain IP addresses I would have though it would block the connection totally and I would not be able to connect at all. This allows the connection and then disconnects anywhere from 1 minute to hours later.
The second change was to add a new printer to the SCO server but all was working fine for a week after this printer was added and I don't see how adding a network printer would have affected the Telnet connections but I am not a SCO Admin to say the least. This is baffling me.
The SCO server is over 20 years old and I am wondering if the NIC card is acting up after all of these years?
I am looking at the wireshark traces to see which end is definitely sending the disconnect. Would anyone know if I should be looking for a disconnect from the Telnet data payload or is it just a matter of the TCP Reset (RST) or Finished (FIN) Flags? Or am I looking too deep for a simple problem?
Thank you all for your contributions.

In your original post you said that your system was release 6.
SCO had a version 5.0.6 issued about 1998, and a release 6.0.0 issued in 2006.
The command "uname -X" will display the release details.
Can you also post (partial) contents of /usr/adm/messages, showing a portion of the file that shows the hardware installed. This will be the same information that shows on the screen when the system boots.
Also, can you run "custom" and take a screen shot of the software installed, to confirm if all patches have been applied.

In 2 similar setups with 5.0.5 I restricted the DHCP pool used by the wireless to something like 10 possible addresses then went static with everything else.

I could see the disconnects as a possible router issue when using DHCP assigned addresses.

I've used TinyTerm over the years, never have had any issues in the 8 installations it was installed in.

jgt

 # uname -X
 System = SCO_SV
Node = service
Release = 5v6.0.0
KernelID = 2008-12-09
Machine = Pent III
BusType = ISA
Serial = 3OC000888
Users = 20
OEM# = 0
Origin# = 1
NumCPU = 4
 #

 

 custom
 IMMMMMMMMMMMMMMMMMMM Software Manager (custom) on service MMMMMMMMMMMMMMMMMMM;:ZDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDD?::3 Host   Software   View   Options                                    Help 3::@DDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDY::                                                                            :: All software on service.AirIdeal.com                                       ::ZDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDD?::3*       BackupEDGE for OpenServer 6 (ver 02.03.01)                        3::3        SCO OpenServer Release 6.0.0 (ver 6.0.0Ni)                        3::3        SCO OpenServer Release 6.0.0 Maintenance Pack 4 (ver 1.0.0Ce)     3::3                                                                          3::3                                                                          3::3                                                                          3::3                                                                          3::3                                                                          3::3                                                                          3::3                                                                          3::3                                                                          3::3                                                                          3::3                                                                          3::@<************************************************DDDDDDDDDDDDDDDDDDDDDDDD>Y:: View All                                             Total 3   Selected 1  ::DDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDD:: The selected software is fully installed.                                  :HMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM<

 

I have searched the /usr/adm/messages file and do not see the hardware except for the very beginning of the file which is from years ago before I started,

 WARNING: Integral console keyboard not found. If you are using
the integral console, check the keyboard connection.
Wed Aug  5 10:26:31 2009
 WARNING: svc_start: ss_accept 4
 Aug  5 10:26:31 rpcbind: Terminating on signal. Restart with rpcbind 
Wed Aug  5 10:29:22 2009
ing up.  Please wait.
 
NOTICE: megasas: 00.00.03.05 Mon Oct 02 11:21:32 PDT 2006
0:0,127,0: HBA          : (megasas,1) LSI MegaRAID3.01
  0,8,0: SES          : DP      BACKPLANE       1.05
  1,127,0: HBA          : (megasas,1) LSI MegaRAID3.01
  2,127,0: HBA          : (megasas,1) LSI MegaRAID3.01
  2,0,0: DISK         : DELL    PERC 5/i Adapter1.03
  3,127,0: HBA          : (megasas,1) LSI MegaRAID3.01
  V3.04 (adpu320,1) Adaptec Ultra320 SCSI: Bus 5, Device 12:0,7,0: HBA          : (adpu320,1) Adaptec Ultra320
  0,2,0: TAPE         : SEAGATE DAT    DAT72-052A16E
  V3.04 (adpu320,2) Adaptec Ultra320 SCSI: Bus 5, Device 13:0,7,0: HBA          : (adpu320,2) Adaptec Ultra320
 NOTICE: IDE: DMA enabled(Mode = ULTRA DMA 5) for Drive0 at 0xAC98
 1:0,2,0: HBA          : (ide,1) Generic IDE/ATAPI
  0,0,0: CDROM        : HL-DT-STDVD-ROM DH10N   0D09
  1,2,0: HBA          : (ide,1) Generic IDE/ATAPI
 NOTICE: IPsec: Initialized Security Association Processing.
Aug  5 10:29:22 sco_pmd[982]: PMD started - PID 983
Aug  5 10:29:22 sco_pmd[983]:     Please register your SCO software products now by accessing
Aug  5 10:29:22 sco_pmd[983]:     Xinuos Inc. | Support | Xinuos Product Registration.
Aug  5 10:29:22 sco_pmd[983]:     SCO Product Registration provides you with many important
Aug  5 10:29:22 sco_pmd[983]:     benefits, including access to Technical Support (if you have
Aug  5 10:29:22 sco_pmd[983]:     purchased a license with bundled support) and information on new
Aug  5 10:29:22 sco_pmd[983]:     maintenance, products, services and special promotions.
Aug  5 10:29:22 sco_pmd[983]:     Registering software products is also an important part of
Aug  5 10:29:22 sco_pmd[983]:     controlling software piracy, which helps reduce overall costs
Aug  5 10:29:22 sco_pmd[983]:     of software products.
Aug  5 10:29:22 sco_pmd[983]:     You may use SCO's online help facility for information on how
Aug  5 10:29:22 sco_pmd[983]:     to use the License Manager utility to register your software products.
device    address vec dma  comment
-------------------------------------------------------------------------------
%bcme0    -   22   -  chip=BCM5721 mem=FE3F0000 addr=00:22:19:d7:56:66
 NOTICE: bcme0 (slot:0 port:1): NIC Link is Up 100Mbps half duplex
synchutmp (Aug  5 10:29:27 pid/ppid 1111/1): truncating /var/adm/wtmpx to an exact number of records 5572-364
Wed Aug  5 10:29:29 2009
 

is there a way to get this information from today?
The NIC card still has the Same MAC address.

I also see the following information from the messages file

 -------------------------------------------------------------------------------
%bcme0    -   16   -  chip=BCM5721 mem=FE3F0000 addr=00:22:19:d7:56:66
Aug  6 09:43:35 prngd[1117]: prngd 0.9.29 (12 Jul 2004) started up for user root
Aug  6 09:43:35 prngd[1117]: have 7 out of 2048 filedescriptors open
 Aug  6 09:43:37 NOTICE: bcme0 (slot:0 port:1): NIC Link is Up 100Mbps full duplex, receive  & transmit flow control ON
 UX:psradm: INFO: At 08/06/15 09:43:38, 1 was brought on-line.
UX:psradm: INFO: At 08/06/15 09:43:40, 2 was brought on-line.
 ex, receive  & transmit flow control ON
                     UX:psradm: INFO: At 08/06/15 09:43:42, 3 was brought on-line.
 ex, receive  & transmit flow control ON
                                                                                   Aug  6 09:44:13 prngd[1629]: prngd 0.9.29 (12 Jul 2004) started up for user root
Aug  6 09:44:13 prngd[1629]: have 7 out of 2048 filedescriptors open
Aug  6 09:44:13 sshd[1699]: error: Bind to port 22 on :: failed: Network is unreachable.
Aug  6 09:44:13 sshd[1699]: Server listening on 0.0.0.0 port 22.
 

---------- Post updated at 10:57 AM ---------- Previous update was at 10:50 AM ----------

sorry about the "custom" formatting.
Here it is again

 IMMMMMMMMMMMMMMMMMMM Software Manager (custom) on service MMMMMMMMMMMMMMMMMMM;
 :ZDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDD?
 ::3 Host   Software   View   Options                                    Help 3:
 :@DDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDY:
 :                                                                            :
 : All software on service.xxxxxxxxx.com                                       :
 :ZDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDD?:
 :3*       BackupEDGE for OpenServer 6 (ver 02.03.01)                        3:
 :3        SCO OpenServer Release 6.0.0 (ver 6.0.0Ni)                        3:
 :3        SCO OpenServer Release 6.0.0 Maintenance Pack 4 (ver 1.0.0Ce)     3:
 :3                                                                          3:
 :3                                                                          3:
 :3                                                                          3:
 :3                                                                          3:
 :3                                                                          3:
 :3                                                                          3:
 :3                                                                          3:
 :3                                                                          3:
 :3                                                                          3:
 :3                                                                          3:
 :@<************************************************DDDDDDDDDDDDDDDDDDDDDDDD>Y:
 : View All                                             Total 3   Selected 1  :
 :DDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDD:
 : The selected software is fully installed.                                  :
 HMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM<

 

---------- Post updated at 11:00 AM ---------- Previous update was at 10:57 AM ----------

Thank you edfair,
I have a separate DHCP Server running on the Cable router providing addresses for all of the devices, wired and wireless. Are you suggesting setting up a DHCP range for the wireless devices and setting the wired device which require Telnet access to static?

Earlier versions of the OS created and appended the hardware discovery information to the messages file on every boot. I suspect that this is an issue you should look at some time in the future as it is the start of most diagnostic routines. (I assume you have rebooted since 2009)

I had mentioned going static with the terminals as something I had done in my installations. It makes network diagnostics somewhat easier.

@Ed, version 6.0.0 does not display the same startup screen as Openserver 5, and that same information is not recorded in /usr/adm/messages.
Sean, you should download and install some additional patches.
OSS711E, OSS714A, P535239A,P535243B.
These are all available at Xinuos, Inc. | Support | Update | Download | Product
There may be others that are also useful (network drivers, hba.)

jgt
Thank you. I will look into these. I just wonder why after having worked for about a month why it stopped and started disconnecting randomly. Being in IT I know anything, especially a bug in software and rear it ugly head at anytime, but it still perplexes me.
I do see several TCP retransmissions and the Users computers send a TCP RST in wireshark just before every session is disconnected, and so maybe new drivers and updates will help. I will have to check with the Developer that created the proprietary database software as I don't want to completely stop what access they have now. Working on a 20 year old machine and hard drives causes anxiety. I hope that will correct the issue, I should know in a couple of days and will post progress.
All of the help and suggestions are greatly appreciated

To all.
I still need your help on this one.
My apologies on the delay, but I needed to get approval to load the SCO fixes as listed by jgt but this was completed on Friday August 21st and as of today the issue happened again. My user was connected via telnet into the server working in the database application an suddenly they just get disconnected. Depending on what terminal emulation program they are using they see "telnet session terminated" on the screen or the entire screen just disappears. Again this depends on how the terminal emulation program responds to being disconnected. Any help, ideas or suggestions would be greatly appreciated.

Can you try the following demontstration.

#cd /tmp
#l t
***you should get a error message saying that t does not exist.
#cat >t <<EOF
echo Goodbye World
EOF
#chmod +x t
#./t
  **you should see "Goodbye World"
#exec ./t

The telnet session disappears.
Can you determine how the application code is started?

#grep username /etc/password

The output shows the user, uid,groupid,name, home directory, and shell. Separated by colons.
Is the home directory the same for all users.
Is the shell, a common shell (sh,ksh,bash,rksh) or is it the application program?
If it is not the application program, what are the last several lines of the user's .profile (this is a file called .profile in the user's home directory.)
Before trying anything we should know if all users share a common home directory.

#cd /etc
#less passwd
you can use <pgdn> and <pgup> to view, and type q at the : prompt to quit.