limit spiders to 50, not 500 to prevent oom.
resume killed merges that had num files shrunk even
if down to one file. show collnum in spider queue.
remove back-to-back whitespace, and make all space
a ' ' for getting the doc checksum for deduping.
"lastspidertime>={roundstart} --> IGNORE"
so we can spider all urls before we advance
to the next spider round and re-spider everything
again. CollectionRec::m_spiderRoundStartTime and
CollectionRec::m_spiderRoundNum are the new
collection rec parms. show the round stuff
on url filters page.
files to gigablast. it should index
just those dmoz urls, and not spider their links.
it should ignore external errors like
ETCPTIMEDOUT when indexing so it will be
identical to dmoz.
query reindex recs can just use their
prob docids as fake uh48s. we need it so we
can avoid the fakedb record and just use
the spider reply to trigger a 5-second
lock expiration. a little simpler. added
logdebugspiderwait for waiting tree debugging.
fixed per ip spider limiting. fixed losing
spiders down blackhole from updateCrawlInfo.
check UrlLock::m_confirmed when counting outstanding
spiders on one ip since may have a lock on one host
but not get granted on all! it calls
confirmLockAcquisition() when it gets fully granted
the lock so it can set UrlLock::confirmed.
fix maxspidersperip limitation not being obeyed.
removed fakedb.
only add "0" time waiting tree keys to waiting tree.
only scanSpiderdb() will change their times to
a future time or add them to doledb directly.
confirmLockAcquisition() will not add to waitingtree
if max spiders per ip limit would be exceeded.
an incoming spider reply will trigger the add to
waiting tree with a time of "0".
spider request so urldata=xxxx can turn link
spidering off. probably desirable so its default.
so &spiderlinks=[0|1] applies to urldata as well
as injecturl=
use msg12 to remove rec from doledb/doleiptable
and add 0 entry to waiting table so doledb is
again immediately repopulated with that firstIp
so we can spider multiple urls from the same ip
at the same time.
collections, really limited by the sizeof(collnum_t) only now,
which is 16bits, 15bits unsigned, which is the limitation.
can always expand this so we can have more than 32k collections.
doledb at the very first key in spiderdb.
causes lots of positive/negative key annihilations.
we end up re-reading like 300 times in some
cases just to get a url from a doledb priority.