Re: [OpenCALEA] Recent updates

[Jesse Norell]

> I verified that
> If lea_collector is not running, df_collector saves the files CmC and CmII
> in the corresponding subfolders.
> If lea_collector is running, CmC files are saved but remain empty.

  That's actually the intended behavior for what was done there ... it's
not actually in the ATIS spec, but Norm wanted to "error on the side of
caution" here and save data to disk if the writes to CF (lea_collector)
failed.  The standard is remiss in how to handle any type of
retransmission/errors/failures/etc.  Perhaps we don't want an empty file
created when all writes succeed, though it's mostly harmless.


> Just one question, in the new design, how can I make the df_collector to
> connect to a remote host running the lea_collector?
> 
> In the old design, tap was connecting to the lea_collector (no problem)
> But now, df_collector connects to localhost, which is the default behavior.
> The df_collector usage says:
> Usage: df_collector  [-f config-file] [-u user] [-g group] [-b bind-addr]
> [-p df-port]  [-v [...]] [-D debug-file] [-l log-level ] [-L logfile] [-h]
> 
> It seems that there is no option to connect to a remote host, or as you said
> "the docs don't often line up  with what's currently possible."

  Correct .. to handle multiple intercepts and also multiple
destinations of a single intercept (eg. part of T1.678 allows the
content of each separate voice channel to be sent to a different port on
the CF host, which would have to happen dynamically), we introduced
"routes" to df_collector.  Ie. this is for picking the destination of
where something (CmII or CmC) goes, not the OS network routing table.
You create a route in df_collector, say to a given host/port/protocol,
then when you send a packet of CmII or CmC it has a routeid which is
used to look up the destination.  We'll be using the same functionality
to "route" to (ie. save to) a file soon.  In the short term it's just
pieced together - tap sends an add route message to df_collector, I
think it's hard-coded as to the destination.  Long term (eg. once the
control daemon is in place, which is coming along) the route will
normally be created when an intercept is defined/started, and tap will
be told what routeid to stick in it's packets.

-- 
Jesse Norell - jesse@kci.net
Kentec Communications, Inc.