Fatal IO error: client killed. Silent crash and this is the only hint I have.
-
I have an application, running on CentOS 6.4. It isn't particularly complex, just displays some data on a GUI with a few pixmaps here and there. It seems to run fine for hours on end before crashing but the length of time varies greatly (sometimes crashes after 4 hours, sometimes 16). The only hint I have for why it crashes is that "Fatal IO error: client killed" gets output. In all the times it has crashed, it has only produced a single core file (relevant section appended below).
Can anyone give me a hint as to what might be causing this? Or at least a good starting point as to how I might catch this error?
@#0 0x00392424 in __kernel_vsyscall ()
#1 0x0077db11 in raise () from /lib/libc.so.6
#2 0x0077f3ea in abort () from /lib/libc.so.6
#3 0x00776deb in __assert_fail_base () from /lib/libc.so.6
#4 0x00776ea6 in __assert_fail () from /lib/libc.so.6
#5 0x092e3a6e in _XAllocID () from /usr/lib/libX11.so.6
#6 0x092c13f8 in XCreateGC () from /usr/lib/libX11.so.6
#7 0x00da63ef in ?? () from /lbt/tcs_runtime/qt-4.7.3/lib/libQtGui.so.4
#8 0x00cd6294 in QPainter::begin(QPaintDevice*) () from /lbt/tcs_runtime/qt-4.7.3/lib/libQtGui.so.4
#9 0x01051f7e in QPushButton::paintEvent(QPaintEvent*) () from /lbt/tcs_runtime/qt-4.7.3/lib/libQtGui.so.4
#10 0x00bc12e4 in QWidget::event(QEvent*) () from /lbt/tcs_runtime/qt-4.7.3/lib/libQtGui.so.4
#11 0x00f9ffd3 in QAbstractButton::event(QEvent*) () from /lbt/tcs_runtime/qt-4.7.3/lib/libQtGui.so.4
#12 0x01050e38 in QPushButton::event(QEvent*) () from /lbt/tcs_runtime/qt-4.7.3/lib/libQtGui.so.4
#13 0x00b5e2ef in QApplicationPrivate::notify_helper(QObject*, QEvent*) ()
from /lbt/tcs_runtime/qt-4.7.3/lib/libQtGui.so.4
#14 0x00b6587e in QApplication::notify(QObject*, QEvent*) ()
from /lbt/tcs_runtime/qt-4.7.3/lib/libQtGui.so.4
#15 0x0050e1fb in QCoreApplication::notifyInternal(QObject*, QEvent*) ()
from /lbt/tcs_runtime/qt-4.7.3/lib/libQtCore.so.4
#16 0x00bbf8d0 in QWidgetPrivate::drawWidget(QPaintDevice*, QRegion const&, QPoint const&, int, QPainter*, QWidgetBackingStore*) () from /lbt/tcs_runtime/qt-4.7.3/lib/libQtGui.so.4
#17 0x00bc03de in QWidgetPrivate::paintSiblingsRecursive(QPaintDevice*, QList<QObject*> const&, int, QRegion const&, QPoint const&, int, QPainter*, QWidgetBackingStore*) ()
from /lbt/tcs_runtime/qt-4.7.3/lib/libQtGui.so.4
@ -
It is steady throughout the run of the app when we're watching it. However, since it can run for hours without any problems, and we have no indication of why it fails, we haven't seen it fail. We only see it has failed after the fact. Also, if it is leaking memory, I think that would show up in a log somewhere, correct? We haven't seen anything to indicate memory is the issue.
-
Hi and welcome to devnet,
Can you try with a more recent version of Qt ? 4.7.3 is rather old
-
I didn't mean that you should replace it right away, just if you can test it with a new version in your test environment to see if it still happens. If not then there might have been a bug that's been fixed in between
-
I'll keep it in mind for eventual test scenarios if what I have in mind right now proves fruitless. Version changes would be a huge deal for us.
For additional info, and probably something I should have included before, we run this on an Xvnc display for remote viewing. We start an XVnc server and then start our app on a vnc display. Is it possible that an issue between my app and the Xvnc server is causing this?
-
That's a possibility yes, maybe network glitches… Did you took a look at the "bug report system":http://bugreports.qt-project.org to see whether there's something about it ?
-
This "one":https://bugreports.qt-project.org/browse/QTBUG-17888 was the closest one, but there is no resolution and I am not seeing any resource leakage.
-
Just a crazy idea… Do you have a screen saver that is triggered ? These might have unsuspected side effects
-
i am facing a similar issue with a Qt3 application. Have you ever found out what the problem was?
-
I have finally found the reason for the issue, its a sequence number wrap-around bug in the libX11: http://stackoverflow.com/questions/23871516/xreply-terminates-app-with-xioerror
