[Xastir-Dev] TODO for "stable"?

Gerry Creager N5JXS gerry.creager at tamu.edu
Wed May 14 20:44:35 EDT 2003


Curt Mills, WE7U wrote:
> On Mon, 12 May 2003, Gerry Creager N5JXS wrote:
> 
> 
>>Curt Mills, WE7U wrote:
>>
>>>Can you definitely pin this on Xastir, or could it perhaps be
>>>changes in the server on the other end, or partial loss of network
>>>connectivity?
>>
>>There is partial connectivity loss.  Every 2-4 hours (I didn't set up
>>the ISDN modem pool, the guy who did failed to document and then left,
>>and the new guy and I haven't figured out the damage yet) I get dumped
>>and have to reacquire a DHCP lease with a new address.
>>
>>In rev's past, it'd time out on xastir and if it was configured to
>>reconnect on network failure, it'd do so, gracefully.  Now, it appears
>>that it "reconnects" and then no more data is exchanged.  Usually takes
>>an uptime of 24+ hours.
>>
>>I can fix it with a cron restart every midnight if you'd like!
> 
> 
> You know, that's a reasonable solution for some closed-source
> programs (or operating-systems), but probably not for us.  ;-)

:-)

> Try out the latest code and see if it's any better in this regard.
> I did some CPU-usage fixes night-before-last, and it might have
> free'd up some CPU for doing other checks.  Then again, perhaps not.

lots of freed cpu cycles, but it locked up last night.  Responded to a 
killall, but it died in a screen refresh.  Didn't have logging or 
verbose on.  Sorry.  That's fixed, now, with the latest (30 minutes old) 
cvs update and recompile.

> The thing I'm thinking about here is that the receive queue can grow
> quite a bit with the older code (older being two days old here), and
> perhaps the networking code didn't recognize that things were down
> 'cuz it was still processing packets?  A bit of a long-shot there,
> but it's possible.

Seen something similar on Mesonet processing data.  Makes sense.

> The network checks every 7.5 minutes haven't been changed for quite
> a while, near as I can tell.  I don't think anyone's been into that
> part of the code.

Thanks for checking.  I don't recall it being changed, either, but one 
never knows.

gerry
-- 
Gerry Creager -- gerry.creager at tamu.edu
Network Engineering -- AATLT, Texas A&M University	
Cell: 979.229.5301 Office: 979.458.4020 FAX: 979.847.8578
Page: 979.228.0173
Office: 903A Eller Bldg, TAMU, College Station, TX 77843



More information about the Xastir-dev mailing list