BIND 10 #1172: Stats show with multiple items crashes bindctl
BIND 10 Development
do-not-reply at isc.org
Thu Mar 22 09:30:59 UTC 2012
#1172: Stats show with multiple items crashes bindctl
-------------------------------------+-------------------------------------
Reporter: jreed | Owner: jelte
Type: | Status: reviewing
defect | Milestone:
Priority: | Sprint-20120403
medium | Resolution:
Component: bind- | Sensitive: 0
ctl | Sub-Project: Core
Keywords: | Estimated Difficulty: 4
Defect Severity: | Total Hours: 0
Medium |
Feature Depending on Ticket: |
Add Hours to Ticket: 0 |
Internal?: 0 |
-------------------------------------+-------------------------------------
Changes (by vorner):
* owner: vorner => jelte
Comment:
Hello
Replying to [comment:8 jelte]:
> I couldn't reproduce this specific error; I did not get a crash, but an
error message.
>
> I could find one way to crash it in the interpreter, and have fixed that
(first commit).
So, congratulation for fixing a different issue than this ticket
described. If I were a bureaucrat, I'd say you should have closed this
ticket and created a new one instead, putting it to next sprint proposed,
etc, etc…̣ :-P But as the new problem is solved, I think it is OK to keep.
Anyway, I believe it can't be reproduced because the Stats show got an
extra argument recently and it acts differently. I can reproduce with very
similar backtrace here:
{{{
> StatsHttpd status x y z
Traceback (most recent call last):
File "/usr/lib64/python3.2/cmd.py", line 212, in onecmd
func = getattr(self, 'do_' + cmd)
AttributeError: 'BindCmdInterpreter' object has no attribute
'do_StatsHttpd'
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "./bin/bindctl", line 149, in <module>
result = tool.run()
File "/home/vorner/testing/bind10/lib64/python3.2/site-
packages/bindctl/bindcmd.py", line 138, in run
self.cmdloop()
File "/usr/lib64/python3.2/cmd.py", line 138, in cmdloop
stop = self.onecmd(line)
File "/home/vorner/testing/bind10/lib64/python3.2/site-
packages/bindctl/bindcmd.py", line 468, in onecmd
Cmd.onecmd(self, line)
File "/usr/lib64/python3.2/cmd.py", line 214, in onecmd
return self.default(line)
File "/home/vorner/testing/bind10/lib64/python3.2/site-
packages/bindctl/bindcmd.py", line 434, in default
self._parse_cmd(line)
File "/home/vorner/testing/bind10/lib64/python3.2/site-
packages/bindctl/bindcmd.py", line 570, in _parse_cmd
self._validate_cmd(cmd)
File "/home/vorner/testing/bind10/lib64/python3.2/site-
packages/bindctl/bindcmd.py", line 368, in _validate_cmd
params[param_count - 2] += params[param_count - 1]
KeyError: -1
}}}
> I know this could be considered out of scope usability stuff, but it was
relatively easy output-only fixes that i think are very useful.
OK, that looks nice. We need more of these output fixes, they make the
software more sexy ;-).
> Not sure if this needs a changelog; if so
> [bug] Fixed a parser bug in bindctl that could make bindctl crash. Also
improved 'command help' output; argument order is now shown correctly, and
parameter descriptions are shown as well.
Maybe it'll be needed. I believe it can be the same after ↑ is fixed.
Replying to [comment:9 jreed]:
> I also can no longer reproduce my crash issues related to this using
recent version. (I am not sure why I said stats remove also had that issue
as that command does not exist.)
That command no longer exists, as it was removed recently. But it did
exist. And I believe this is a problem of any command with only mandatory
parameters.
--
Ticket URL: <http://bind10.isc.org/ticket/1172#comment:11>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development
More information about the bind10-tickets
mailing list