From: Matthew W. S. Bell <men...@engelbert.(none)> When a child dies off, it seems the EBADF from the pipe to the child is often processed before the SIGCHLD. Therefore, dumping into the debugger is not helpful. It seems resources are tidied up appropriately anyway.
Signed-off-by: Matthew W. S. Bell <matt...@bells23.org.uk> --- src/netconf/core.py | 6 +++--- 1 files changed, 3 insertions(+), 3 deletions(-) diff --git a/src/netconf/core.py b/src/netconf/core.py index 981f372..e00d0da 100644 --- a/src/netconf/core.py +++ b/src/netconf/core.py @@ -145,7 +145,7 @@ class Core(object, logger.Mixin): def _sighandler(self, signal, stack): for key, pipe in self._signal_listening_pipes.iteritems(): if key[0] == signal: - self._log.debug('received signal %s, telling via FD %d...' + self._log.debug('received signal %s, notifying via FD %d...' % (str(signal), pipe.readfd)) pipe.write(str(signal) + '\n\n') @@ -219,8 +219,8 @@ class Core(object, logger.Mixin): try: select.select([fd],[],[],0) except: - self._log.error('FD %d still registered but invalid, removing...' % fd) - import pdb; pdb.set_trace() + self._log.info('FD %d still registered but invalid, removing...' % fd) + #import pdb; pdb.set_trace() del self._fdrreactors[fd] continue raise -- 1.6.2.4 _______________________________________________ netconf-devel mailing list netconf-devel@lists.alioth.debian.org http://lists.alioth.debian.org/mailman/listinfo/netconf-devel