Help with strange bzr connection hangs

[Martin Pool]

On 4 October 2011 06:19, Eli Zaretskii <eliz at gnu.org> wrote:
>> Date: Mon, 03 Oct 2011 19:19:48 +0200
>> From: Eli Zaretskii <eliz at gnu.org>
>>
>> If I use -Dhpss, I see it always stuck at this point in .bzr.log:
>>
>>   8.391  Using fetch logic to copy between RemoteRepository(bzr+ssh://eliz@bzr.savannah.gnu.org/emacs/.bzr/)(RemoteRepositoryFormat(_network_name='Bazaar repository format 2a (needs bzr 1.16 or later)\n')) and CHKInventoryRepository('file:///D:/gnu/bzr/emacs/.bzr/repository/')(RepositoryFormat2a())
>>   8.391  fetching: <SearchResult search:(set(['monnier at iro.umontreal.ca-20111003164956-3dygzh7jbf2k7p9y']), ['rgm at gnu.org-20110820221427-6vbkv2xevrd471fe', 'schwab at linux-m68k.org-20110919211547-bpacfe5tvruisv4p', 'eggert at cs.ucla.edu-20110930202201-vgrjezzuzvexthki', 'cyd at stupidchicken.com-20111001184722-kxn1g5qh5mt3a20a', 'cyd at stupidchicken.com-20110830223807-hqjgxlax2rj3zedr', ...], 8)>
>>   8.406  hpss call w/body: 'Repository.get_stream_1.19', 'emacs/', 'Bazaar repository format 2a (needs bzr 1.16 or later)\n' ('search\nmonnier at iro.u'...)
>>   8.406                2310 bytes
>>   8.672     result:   ('ok',)
>>
>> The full log up to this point is at the end of this message.
>>
>> Sometimes, the problem disappears after some time, sometimes it only
>> disappears after a reboot (or maybe I get impatient and reboot before
>> it disappears).  Lately, a day or two after a system restart, it
>> starts happening again.
>>
>> Other communications, including Web browser SSH to the same host, work
>> fine, even if these hangs happen.
>
> Additional info: setting BZR_SSH in the environment to point to
> plink.exe (instead of using the default paramiko) solves the problem
> and allows "bzr up" to run to completion, albeit slower.  (That using
> plink is slower than paramiko is something I know for quite some time,
> so the slower speed not a surprise.  That it succeeded where paramiko
> failed is a surprise.)
>
> It is, of course, possible that the connection speed is what triggers
> the problem, and that paramiko has nothing to do with this except that
> it facilitates speedier traffic.

I don't know, based on that, what it would be, but the general kind of
thing I would try to find out next in this type of situation is just
what is going on when it is hanging: what is bzr doing, what is the
external ssh transport (if any) doing, and what is the OS tcp socket
doing?  I know more about how to do that on Linux than on Windows, but
generally:

 * pop into the bzr debugger with ctrl-break and then get a backtrace
(type 'bt')
 * is there a windows equivalent to 'netstat -ponet' that shows the
socket state?

m