Hello,
We have a SCO box running OpenServer 5.0.5. This has our business software on it (15 users connecting via terminal emulator (ansio lite)) which has been running for about 12-15yrs strong!
Within the past 6-8 months, at random times of the day, we've been loosing connection to this box. Sometimes once a week, sometimes several times a day! We'd have to reboot the UNIX box and we can continue as usual.
Any ideas as to what could be causing this box to go down?
I don't think it's the network card because I have built an identical box and genned the software onto it and still have the same issues.
Thanks.
Ralph C.
Any messages like duplicate ip somewhere else?
What happens if you restart tcp - /etc/tcp stop; /etc/tcp start ?
Could still be a network card on one of the clients.
Run
#crash
>strstat |more
....
>quit
#
This will tell you if there have been any packet errors.
Are all terminal sessions affected at the same time?
Do you get a response using ftp, or http. (ie is tcp down, or just telnet)
I don't pretend to be a UNIX guru, unfortunately. But next time it goes down, I will stop/start and see what results I get.
I will def be in touch!
Thank you so much 
Ralph
---------- Post updated at 09:32 AM ---------- Previous update was at 09:11 AM ----------
Well, everyone was booted from thier term emulator sessions again. While it was down I shutdown and restarted tcp from the console and we were still not able to login.
I also ran '#crash' --- >strstat |more and got a report.
I don't know how to retrieve it though, sorry.
I've also gotten data from typing netstat -im, -in, -rn
It states: SIOCGIFALL: out of stream resourse, as well as 'class 6, fail 290.
Don't know what it all means.
Please help.
Ralph
Does anyone know what, "out of stream memory" means?
Thanks,
RPC
Please post exactly what you typed. This is a syntax error:
netstat -im, -in, -rn
Please then post the exact reply from the command(s) with no typing errors.
If you saw a message "out of stream memory", where did you see that message? It doesn't look like a SCO unix error message.
There is a common message "Out of stream resources" which can cause these symptoms. Try google for your exact error message.
Or at SCO.
http://wdb1.sco.com/kb/search
Out of streams memory is a configuration problem.
Run scoadmin->Kernel Hardware Mgr-->Tune Parameters
Select option 12
Bump NSTRPAGES, try a value twice what is there now.
Have you added users, made the application software more complex, added new applications since this problem began?
Run "sar -A" and see if you are running out of memory (real + swap)
We have not done a thing to this box in years! Maybe that's part of the problem? We have the same users, etc.
When i run 'sar -A' I get 'can't open /usr/adm/sa/sa14'
As for the netstat command, it is 3 seperate commands
netstat -im
netstat -in
netstat -rn
when I run netstat -im from the console, after we have lost connection to the box and before I reboot it (init 6) I get a report that gives a '756' in the failed column and then states, 'total configured streams memory 2000.00kb. Streams memory in use : 1901.43kb. max streams memory used 2116.30kb'
netstat -in produces the following:
netstat: get_if_all: SIOCGIfall: out of stream resourses
As I think I've state before, there have been no changes to this SCO box, only to our network. We are still running Windows Workgroups (no server), but have upgraded to gigabit switches and a new Cisco ASA firewall. We've changed to optonline as our ISP and now have a VOip phone system.
Thanks for your help!!
Ralph C.
Uncomment the lines in /usr/spool/cron/crontabs/sys and /usr/spool/cron/crontabs/root that refer to sa and sa2.
Performance data will be kept in /usr/sys/admin/sa/sa01 through sa31 for a rotating month. Data is collected every 20 minutes.