[Fab-user] timeouts

[Michael Bravo]

Hello,

I'd like to raise the issue of timeouts. There is at least one case I
run into repeatedly and quite reliably. It occurs on a task I run
across a number of database hosts, where a set of tables gets dumped,
compressed and then transferred. On some of those hosts the dumps are
quite extensive, and running "gzip -9" on them takes considerable
time, more than a dozen minutes. And it's where the problem occurs.
>From time to time I am forced to ssh to the host in question
separately and observe that the gzip process is not running anymore
(having completed its job), while Fabric is not aware of it and just
sits there, never timing out. It also results in the necessity for me
to interrupt or kill the Fabric process and, depending on the position
of the host in question on the list, make manual readjustments to
complete the task across the host list. It's rather inconvenient. Is
there anything that can be done to prevent this, or some debugging
info I could provide to help pinpoint the problem? To be honest, I'm
not exactly sure where the problem lies, does the actual network
connection time out (seems unlikely), or is something happening with
the actual detection of events on the other side of the connection.

P.S. From browsing through existing issues, I understand that the
problem like that probably lies with Paramiko, would that be correct?
-- 
/\/\ike