Re: [Evolution] I/O operation timing out

[Adam Tauno Williams <awilliam whitemice org>]

On Sat, 2013-11-30 at 11:49 -0500, Eric Beversluis wrote:
> On Mon, 2013-10-28 at 11:00 -0400, Adam Tauno Williams wrote:
> > On Mon, 2013-10-21 at 18:58 -0400, Eric Beversluis wrote:
> > > I now have Evo 3.8.5 on Fedora 19. So everything's up to date in Kansas
> > > City. The problem persists. I'm doing 'send/receive' on three accounts,
> > > all the same domain, all hosted at omnis.com. I've done some traceroutes
> > > and my impression is that most if not all of the jumps take longer when
> > > the attempt times out. This presumably would mean that it's not a
> > > problem at the omnis.com end. So I repeat my earlier query: Is there
> > > some way to tell Evolution to wait longer before it times out? I've
> > > already set net.ipv4.tcp_syn_retries to 7, which is supposed to give
> > > about 90 seconds (per
> > > http://www.sekuda.com/overriding_the_default_linux_kernel_20_second_tcp_socket_connect_timeout).
> > > But I'm getting the timeout error after about 45 seconds and 60 seconds.
> > > And it's still intermittent--sometimes stuff downloads right away;
> > > sometimes only one or two accounts download and then it times out.
> > Is this for an IMAP or POP account?
> > If you run "CAMEL_VERBOSE_DEBUG=1 evolution" do you see anything about
> > the timeout?  
> > I'm not an expert but noodling around in the camel code in EDS I do not
> > see anything that looks like a socket timeout; so I'd guess whatever the
> > default is what the default is.
> Evolution experts/developers out there: Is there anything in Evolution
> code that creates a time out? 

Depends on what you mean by "creates a time out".  The technical answer
in yes - any application that performs I/O [the includes network I/O]
can [and should] raise a time-out exception if an operation times out.

Evolution almost certainly does not *cause* a time-out;  the time-out is
'bubbling up' from the underlying subsystem(s).  Evolution does a *LOT*
of I/O  - it is a powerful application and demanding of the underlying
subsystems - so it may very well get an exception where something else
may not.

Not a bug in Evolution.

> Where is the error message that the I/O
> operation timed out coming from?

Almost certainly from your network stack; and my $$$ would be on your
ISP/customer router.


-- 
Adam Tauno Williams <mailto:awilliam whitemice org> GPG D95ED383
Systems Administrator, Python Developer, LPI / NCLA