T.R | Title | User | Personal Name | Date | Lines |
---|
8557.1 | my customer has similar problem | GYPOS3::EIBL | Reinhard Eibl @FKR, DTN 865-4027 | Mon May 12 1997 12:19 | 21 |
| Hi,
have a call with similar problem symptoms:
as 2100, DUNIX 3.2d-2 and lots of
"X Toolkit Warning: Select failed; error code 22"
what fills up his partition.
They are displaying a 3rd party application to either VXT2000 or to PC's
running windows 3.11 and windows nt. X11 emulation used is reflexion 6.0.
Unfortunately there are constraints which prevent a test with local display.
Customer found the error probability to be higher when there is high load
on the network.
Found no hint in our databases or in the 3.2d-2 patches.
Any idea ?
Was there a solution for the problem in re. 0 ?
Any help is appreciated, thanks in advance.
/Reinhard
|
8557.2 | Vrestore de /usr | MDR01::ARANCHA | MCS Madrid | Tue May 20 1997 12:47 | 9 |
| Hi,
My customer has the same problem.
I solved it. We restore the /usr , after initialized the FS.
The customer has Advfs filesystem for / and /usr.
But I don't know why the filesystem is full.
regards,
Arancha
|
8557.3 | We've seen this and have a suggestion... | DECWET::DIPIETRO | | Tue May 20 1997 18:10 | 27 |
| Our dxadvfs application running under 3.2 occasionally encounters this bug.
In our release notes, we have the following suggestion:
If you get a continuous output of 'select error 22' messages
from the GUI on the controlling terminal, then stop and restart
the GUI with the -sync option. Eg.
# dxadvfs -sync
This should stop the flow of error messages.
Thought we filed a QAR on this one, but can't find a record of it.
Haven't seen this one in Steel.
More info:
The error message: "Warning: Select failed; error code 22"
is coming from the file NextEvent.o in libXt.a.
Our theory is that the select error bug is due to a memory bug that
is exposed only occasionally due to the ordering of X protocol messages
between the GUI and the X-server. You can change the timing/order of
X protocol messages by running the application in 'server synchronous mode'.
This slows the application down but definitely affects the bug; At least
one customer now has been able to stop the message spew when nothing
else worked.
|