[ntp:hackers] I/O

Danny Mayer mayer at gis.net
Tue Feb 15 20:13:55 PST 2005


At 10:46 PM 2/15/2005, David L. Mills wrote:
>Danny,
>
>At least in the first few minutes, Solariba deacon and pogo appear okay as 
>well as Freebie mort. However, Freebie rackety, which is configured like 
>mort and should be running the same (your) code, apparently loops. Start 
>it with two -d's and watch the refclock_gtraw() trace. I'm not sure we are 
>out of the woods on the other machines; I'll let them run a bit.
>
>Dave

I just looked. rackety is not running right now. deacon, pogo and mort seem
to be okay, but what would I know?

watching refclock_gtraw() trace  made my eyes glaze over as I don't know
what I'm looking at. I changed MAXZEROREADS to 1 in the source pool
as you said it was okay to get to the next clock or IP interfaces at that 
point.
I didn't delete the recvbuff for a zero read-count so you get to process that
one.

I could give rackety another run.

Danny




More information about the hackers mailing list