Hi,

O/S: W2K and W2003
Universe 10.1

We keep getting errors logged in the event viewer.
There are 2 messages predominating:

Universe warning. Unable to write to socket. WSA error 10053

and

UniVerse warning: Unable to read from stdin.. WSA error: 0.

These errors keep coming up every 30 seconds or so.

We have no idea why they are materialising.

Any thoughts are much appreciated.

cheers
Terry

Terry wrote:

Terry - usually the event viewer gives the actual program/dll name the
error came from, do you have that - That would help narrow it down.

rgds
Symeon.

I wish!

The one line is all we get.



"Symeon" <symeonb (AT) gmail (DOT) com> wrote

The same server periodically locks up the SB+ clients - I don't know if it
is relevant.
For some reason, the server B+ session manager just freezes up.
It looks like a SB+ client is sending a command (telnet) which causes the
server to go into a loop.
It stays indefinately in this frozen state, but only for SB+ telnet clients.
Uniobject access is uninterupted.
There is no unusual CPU activity and a network packet logger does not show
anything unusual.
The weird thing is that it is a different client workstation each time - if
we systematically shut down the workstations, the one that is causing the
problem gets shut down, and the server unfreezes and the other clients are
able to continue.Alternativey, we reboot the server and that frees it up
until the next freeze.
The common delimiter with the freezes seems to be the time: in the morning
when most users are logging on, or after lunch when they log on again, we
get the freezes. It never seems to happen when there is little activity or
just the 'normal' activity of the application.
Bandwidth monitoring shows that the bandwidth even when all users are
logging on is still well below 60% utilisation.
We have 512 kbs asynchronous between WAN sites and only 4-5 users at each
site. It even occassionaly happens on the LAN.

I doubt that it is the SB+ code that is causing the problem as this code is
installed and fine at over 100 other sites.
The only thing unique about this site is that we are using a new Dell
rackmount Powedge RAID 5 SCSI server.
This customer also uses shaping across the WAN.
We've replaced the Server NIC, tried W2003 and W2K without effect.

I am a bit disappointed at the Telnet session manager - it has very poor
logging abilities for troubleshooting.

Any ideas?

Thanks
Terry





"Symeon" <symeonb (AT) gmail (DOT) com> wrote

Terry,

we get these messages when a telnet session is closed/aborted instead
of the user logging off correctly. Can't help with the freezing issue
other than to say that we don't install SBClient/SBClient Session
Manager on the server.

hth, Dean.

We get this too - but it is a result of the users getting frustrated withthe
non-responsiveness of the server and they kill the telnet session in the
task manager.
They can't get any response from their session to be able to logout
correctly, and resort to the task kill in an attempt to get some response.
This (of course) just exacerbates the problem.


<dean (AT) xponet (DOT) ltd.uk> wrote

You mentioned shaping ... That would be my immediate concern !

Is the app SB+ GUI or enhanced character (green screenish) ?


Symeon.

it's SB+GUI
We just changed the Server (for the 3rd time) and it is still happening.


"Symeon" <symeonb (AT) gmail (DOT) com> wrote

Terry wrote:

hmm - i think you (or your Var) should log this with IBM.