Whenever a subcommand fails for some reason, we shouldn't include
the full args of the command in the error output; in some very common
cases, these args can be very, very large, and this just clogs the
logs with large amounts of unreadable junk. If we can come up with a
cleaner way to represent the arguments then we can add this back, but
right now they're actually less useful than nothing at all since the
make it hard to find the real messages that you care about in the
logs.
Signed-off-by: John Admanski <jadmanski@google.com>
git-svn-id: http://test.kernel.org/svn/autotest/trunk@1225 592f7852-d20e-0410-864c-8624ca9c26a4
diff --git a/server/subcommand.py b/server/subcommand.py
index 26b167e..a01b963 100644
--- a/server/subcommand.py
+++ b/server/subcommand.py
@@ -176,14 +176,14 @@
if not pid:
utils.nuke_pid(self.pid)
print "subcommand failed pid %d" % self.pid
- print "%s(%s)" % (self.func, self.args)
+ print "%s" % (self.func,)
print "timeout after %ds" % timeout
print
return None
if status != 0:
print "subcommand failed pid %d" % pid
- print "%s(%s)" % (self.func, self.args)
+ print "%s" % (self.func,)
print "rc=%d" % status
print
if os.path.exists(self.stderr):
@@ -191,5 +191,5 @@
print line,
print "\n--------------------------------------------\n"
raise AutoservError('Subcommand failed:' +
- '%s(%s) rc=%d' % (self.func, self.args, status))
+ '%s rc=%d' % (self.func, status))
return status