* Throttling improvements on SSU - throttle all transmissions to a peer
when we are retransmitting, not just retransmissions. Also, if
we're already retransmitting to a peer, probabalistically tail drop new
messages targetting that peer, based on the estimated wait time before
transmission.
* Fixed the rounding error in the inbound tunnel drop probability.
* Include a combined send/receive graph (good idea cervantes!)
* Proactively drop inbound tunnel requests probabalistically as the
estimated queue time approaches our limit, rather than letting them all
through up to that limit.
* Process inbound tunnel requests more efficiently
* Proactively drop inbound tunnel requests if the queue before we'd
process it in is too long (dynamically adjusted by cpu load)
* Adjust the tunnel rejection throttle to reject requeusts when we have to
proactively drop too many requests.
* Display the number of pending inbound tunnel join requests on the router
console (as the "handle backlog")
* Include a few more stats in the default set of graphs
* Cut down on the time that we allow a tunnel creation request to sit by
without response, and reject tunnel creation requests that are lagged
locally. Also switch to a bounded FIFO instead of a LIFO
* Threading tweaks for the message handling (thanks bar!)
* Don't add addresses to syndie with blank names (thanks Complication!)
* Further ban clearance
* Fix during the ssu handshake to avoid an unnecessary failure on
packet retransmission (thanks ripple!)
* Fix during the SSU handshake to use the negotiated session key asap,
rather than using the intro key for more than we should (thanks ripple!)
* Fixes to the message reply registry (thanks Complication!)
* More comprehensive syndie banning (for repeated pushes)
* Publish the router's ballpark bandwidth limit (w/in a power of 2), for
testing purposes
* Put a floor back on the capacity threshold, so too many failing peers
won't cause us to pick very bad peers (unless we have very few good
ones)
* Bugfix to cut down on peers using introducers unneessarily (thanks
Complication!)
* Reduced the default streaming lib message size to fit into a single
tunnel message, rather than require 5 tunnel messages to be transferred
without loss before recomposition. This reduces throughput, but should
increase reliability, at least for the time being.
* Misc small bugfixes in the router (thanks all!)
* More tweaking for Syndie's CSS (thanks Doubtful Salmon!)
* Take out the router watchdog's teeth (don't restart on leaseset failure)
* Filter the IRC ping/pong messages, as some clients send unsafe
information in them (thanks aardvax and dust!)
* Substantially reduced the lock contention in the message registry (a
major hotspot that can choke most threads). Also reworked the locking
so we don't need per-message timer events
* No need to have additional per-peer message clearing, as they are
either unregistered individually or expired.
* Include some of the more transient tunnel throttling
* Added a simple purge and ban of syndie authors, shown as the
"Purge and ban" button on the addressbook for authors that are already
on the ignore list. All of their entries and metadata are deleted from
the archive, and the are transparently filtered from any remote
syndication (so no user on the syndie instance will pull any new posts
from them)
* More strict tunnel join throtting when congested
* Try to desync tunnel building near startup (thanks Complication!)
* If we are highly congested, fall back on only querying the floodfill
netDb peers, and only storing to those peers too
* Cleaned up the floodfill-only queries
* Avoid a very strange (unconfirmed) bug that people using the systray's
browser picker dialog could cause by disabling the GUI-based browser
picker.
* Cut down on subsequent streaming lib reset packets transmitted
* Use a larger MTU more often
* Allow netDb searches to query shitlisted peers, as the queries are
indirect.
* Add an option to disable non-floodfill netDb searches (non-floodfill
searches are used by default, but can be disabled by adding
netDb.floodfillOnly=true to the advanced config)
* Made the netDb search load limitations a little less stringent
* Add support for specifying the number of periods to be plotted on the
graphs - e.g. to plot only the last hour of a stat that is averaged at
the 60 second period, add &periodCount=60
* Add support for graphing the event count as well as the average stat
value (done by adding &showEvents=true to the URL). Also supports
hiding the legend (&hideLegend=true), the grid (&hideGrid=true), and
the title (&hideTitle=true).
* Removed an unnecessary arbitrary filter on the profile organizer so we
can pick high capacity and fast peers more appropriately
* Integrate basic hooks for jrobin (http://jrobin.org) into the router
console. Selected stats can be harvested automatically and fed into
in-memory RRD databases, and those databases can be served up either as
PNG images or as RRDtool compatible XML dumps (see oldstats.jsp for
details). A base set of stats are harvested by default, but an
alternate list can be specified by setting the 'stat.summaries' list on
the advanced config. For instance:
stat.summaries=bw.recvRate.60000,bw.sendRate.60000
* HTML tweaking for the general config page (thanks void!)
* Odd NPE fix (thanks Complication!)
* Further stat cleanup
* Keep track of how many peers we are actively trying to communicate with,
beyond those who are just trying to communicate with us.
* Further router tunnel participation throttle revisions to avoid spurious
rejections
* Rate stat display cleanup (thanks ripple!)
* Don't even try to send messages that have been queued too long
* HTML fixes in Syndie to work better with opera (thanks shaklen!)
* Give netDb lookups to floodfill peers more time, as they are much more
likely to succeed (thereby cutting down on the unnecessary netDb
searches outside the floodfill set)
* Fix to the SSU IP detection code so we won't use introducers when we
don't need them (thanks Complication!)
* Add a brief shitlist to i2psnark so it doesn't keep on trying to reach
peers given to it
* Don't let netDb searches wander across too many peers
* Don't use the 1s bandwidth usage in the tunnel participation throttle,
as its too volatile to have much meaning.
* Don't bork if a Syndie post is missing an entry.sml
* More aggressive tunnel throttling as we approach our bandwidth limit,
and throttle based off periods wider than 1 second.
* Included Doubtful Salmon's syndie stylings (thanks!)
2006-02-27 jrandom
* Adjust the jbigi.jar to use the athlon-optimized jbigi on windows/amd64
machines, rather than the generic jbigi (until we have an athlon64
optimized version)
* Switch from the bouncycastle to the gnu-crypto implementation for
SHA256, as benchmarks show a 10-30% speedup.
* Removed some unnecessary object caches
* Don't close i2psnark streams prematurely
* Made the Syndie permalinks in the thread view point to the blog view
* Disabled TCP again (since the live net seems to be doing well w/out it)
* Fix the message time on inbound SSU establishment (thanks zzz!)
* Don't be so aggressive with parallel tunnel creation when a tunnel pool
just starts up
* Rounding calculation cleanup in the stats, and avoid an uncontested
mutex (thanks ripple!)
* SSU handshake cleanup to help force incompatible peers to stop nagging
us by both not giving them an updated reference to us and by dropping
future handshake packets from them.
* Fix to properly profile tunnel joins (thanks Ragnarok, frosk, et al!)
* More aggressive poor-man's PMTU, allowing larger MTUs on less reliable
links
* Further class validator refactorings
* Throttle the outbound SSU establishment queue, so it doesn't fill up the
heap when backlogged (and so that the messages queued up on it don't sit
there forever)
* Further SSU memory cleanup
* Clean up the address regeneration code so it knows when to rebuild the
local info more precisely.
* Throttle the outbound SSU establishment queue, so it doesn't fill up the
heap when backlogged (and so that the messages queued up on it don't sit
there forever)
* Further SSU memory cleanup
* Properly enable TCP this time (oops)
* Deal with multiple form handlers on the same page in the console without
being too annoying (thanks blubb and bd_!)
* Major SSU and router tuning to reduce contention, memory usage, and GC
churn. There are still issues to be worked out, but this should be a
substantial improvement.
* Modified the optional netDb harvester task to support choosing whether
to use (non-anonymous) direct connections or (anonymous) exploratory
tunnels to do the harvesting. Harvesting itself is enabled via the
advanced config "netDb.shouldHarvest=true" (default is false) and the
connection type can be chosen via "netDb.harvestDirectly=false" (default
is false).
* Moved the current net's reseed URL to a different location than where
the old net looks (dev.i2p.net/i2pdb2/ vs .../i2pdb/)
* More aggressively expire inbound messages (on receive, not just on send)
* Add in a hook for breaking backwards compatibility in the SSU wire
protocol directly by including a version as part of the handshake. The
version is currently set to 0, however, so the wire protocol from this
build is compatible with all earlier SSU implementations.
* Increased the number of complete message readers, cutting down
substantially on the delay processing inbound messages.
* Delete the message history file on startup
* Reworked the restart/shutdown display on the console (thanks bd_!)
* Migrate the outbound packets from a central component to the individual
per-peer components, substantially cutting down on lock contention when
dealing with higher degrees.
* Load balance the outbound SSU transfers evenly across peers, rather than
across messages (so peers with few messages won't be starved by peers
with many).
* Reduce the frequency of router info rebuilds (thanks bar!)
