Arno has observed "slow startup" -- a new node takes
several minutes to connect to the network and start
doing searches/downloads/uploads properly. I haven't
observed this, but then I haven't tested as much as
Arno has since v0.6.2.364-STABLE was built.
Anyway, personally I think this should be bumped to
v0.6.3 and maybe a note should be added to a "known
bugs" document that goes out with the v0.6.2 release.
Logged In: NO
Arno agrees. It is only very noticeable when a fresh broker
is started the first time, so we should ask new users for
some patience...
Logged In: YES
user_id=52562
This *might* be the same as "memory usage explosion" bug.
Logged In: YES
user_id=642335
for me that is a 0.6.2 since it is connected to reconnection
problems and possible broker isolation.
We avoid bootpage lookup except for broker's life cycle's first
launch, and we often don't send hellos during the first 15
minutes, and we often only see one single MT being helloed at
during the first hour (which might be offline or who knows?),
and we often don't send list servers during the first 5 or so
minutes either.
So only after these times the broker can start communication
with the network.
We should change that. It seems important for a first-user
impression as well, and we want good reputation, no?
Logged In: YES
user_id=52562
This happens only if your node is using relay server.
Arno has correctly identified the details, and we've
discussed it on the mailing list. (Sorry no link here
because the mail archive at sf.net hasn't updated to include
the discussion yet.)
Logged In: YES
user_id=52562
This thread describes the reason that this happens. A nice
workaround or fix should be easy for v0.6.2:
http://sourceforge.net/mailarchive/forum.php?thread_id=3981526&forum_id=7702
Increasing priority!
Logged In: YES
user_id=52562
I believe Arno is now testing a patch for this.
Logged In: YES
user_id=52562
As of nano 372, this does not happen anymore.
But I am leaving this issue open for Arno to close it after
he tests that patch.