Forgive me for beating this horse again, but I believe the code below
from list.c explains why the update daemon (nntpcache-1.0) waits until
the retry interval has expired, even after a failed attempt to get
active or another list.
#ifdef MMALLOC
if (list->type == l_overview_fmt)
{
if (pid>0)
if (getListServer(list, scfg)>0)
scfg->overview_fmt_rebuild = tim;
} else
if (pid == 0)
#endif
getListServer(list, scfg);
switch (list->type)
{
case l_active:
scfg->active_rebuild = tim;
break;
case l_active_times:
scfg->active_times_rebuild = tim;
break;
case l_newsgroups:
scfg->newsgroups_rebuild = tim;
break;
#ifndef MMALLOC
case l_overview_fmt:
scfg->overview_fmt_rebuild = tim;
break;
#endif
default:
break;
}
Unless I'm misreading this code, for lists other than l_overview_fmt,
the return value of getListServer() is not checked. Hence, even if
the attempt to get the active file fails -- for whatever reason -- the
rebuild time is changed, and the update daemon won't try to get the
active list again until the configured interval (4h here) has expired.
Shouldn't the rebuild time only get changed when the attempt to get
the active list (or another list) succeeds?
--
Ronald Florence Maple Lawn Farm, Stonington, CT
ron@mlfarm.com http://www.connix.com/~mlfarm
