net.i2p.router.networkdb.kademlia
Class ExploreJob

java.lang.Object
  extended bynet.i2p.router.JobImpl
      extended bynet.i2p.router.networkdb.kademlia.SearchJob
          extended bynet.i2p.router.networkdb.kademlia.ExploreJob
All Implemented Interfaces:
Job

class ExploreJob
extends SearchJob

Search for a particular key iteratively until we either find a value, we run out of peers, or the bucket the key belongs in has sufficient values in it. Well, we're skipping the 'bucket gets filled up' test for now, since it'll never get used (at least for a while).


Nested Class Summary
protected  class SearchJob.FailedJob
          Called when a particular peer failed to respond before the timeout was reached, or if the peer could not be contacted at all.
 
Field Summary
protected  KademliaNetworkDatabaseFacade _facade
           
(package private) static int MAX_CLOSEST
          only send the 10 closest "dont tell me about" refs
(package private) static int PER_FLOODFILL_PEER_TIMEOUT
           
 
Constructor Summary
ExploreJob(RouterContext context, KademliaNetworkDatabaseFacade facade, Hash key)
          Create a new search for the routingKey specified
 
Method Summary
(package private)  boolean add(Hash peer)
           
 int addDeferred(Job onFind, Job onFail, long expiration, boolean isLease)
           
protected  DatabaseLookupMessage buildMessage(long expiration)
          We're looking for a router, so lets build the lookup message (no need to tunnel route either, so just have replies sent back to us directly).
protected  DatabaseLookupMessage buildMessage(TunnelId replyTunnelId, Hash replyGateway, long expiration)
          Build the database search message
protected  DatabaseLookupMessage buildMessage(TunnelId replyTunnelId, RouterInfo replyGateway, long expiration)
          Build the database search message, but unlike the normal searches, we're more explicit in what we /dont/ want.
protected  void continueSearch()
          Send a series of searches to the next available peers as selected by the routing table, but making sure no more than SEARCH_BREDTH are outstanding at any time
(package private)  void decrementOutstandingFloodfillSearches()
           
protected  void fail()
          Search totally failed
protected  int getBredth()
          max # of concurrent searches
 long getExpiration()
           
protected  KademliaNetworkDatabaseFacade getFacade()
           
 java.lang.String getName()
          Descriptive name of the task
protected  int getPerPeerTimeoutMs()
          Let each peer take up to the average successful search RTT
protected  int getPerPeerTimeoutMs(Hash peer)
           
protected  SearchState getState()
           
 long getTimeoutMs()
           
(package private) static boolean isCongested(RouterContext ctx)
           
protected  void newPeersFound(int numNewPeers)
          We've gotten a search reply that contained the specified number of peers that we didn't know about before.
(package private) static boolean onlyQueryFloodfillPeers(RouterContext ctx)
           
(package private)  void replyFound(DatabaseSearchReplyMessage message, Hash peer)
           
 void runJob()
          Actually perform the task.
protected  void searchNext()
          Send the next search, or stop if its completed
protected  void sendLeaseSearch(RouterInfo router)
          we're (probably) searching for a LeaseSet, so to be (overly) cautious, we're sending the request out through a tunnel w/ reply back through another tunnel.
protected  void sendRouterSearch(RouterInfo router)
          we're searching for a router, so we can just send direct
protected  void sendSearch(RouterInfo router)
          Send a search to the given peer
(package private)  long timeoutMs()
           
 java.lang.String toString()
           
(package private)  boolean wasAttempted(Hash peer)
           
 
Methods inherited from class net.i2p.router.JobImpl
dropped, getAddedBy, getContext, getJobId, getMadeReadyOn, getTiming, madeReady, requeue
 
Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait
 

Field Detail

_facade

protected KademliaNetworkDatabaseFacade _facade

MAX_CLOSEST

static final int MAX_CLOSEST
only send the 10 closest "dont tell me about" refs

See Also:
Constant Field Values

PER_FLOODFILL_PEER_TIMEOUT

static final int PER_FLOODFILL_PEER_TIMEOUT
See Also:
Constant Field Values
Constructor Detail

ExploreJob

public ExploreJob(RouterContext context,
                  KademliaNetworkDatabaseFacade facade,
                  Hash key)
Create a new search for the routingKey specified

Method Detail

buildMessage

protected DatabaseLookupMessage buildMessage(TunnelId replyTunnelId,
                                             RouterInfo replyGateway,
                                             long expiration)
Build the database search message, but unlike the normal searches, we're more explicit in what we /dont/ want. We don't just ask them to ignore the peers we've already searched on, but to ignore a number of the peers we already know about (in the target key's bucket) as well. Perhaps we may want to ignore other keys too, such as the ones in nearby buckets, but we probably don't want the dontIncludePeers set to get too massive (aka sending the entire routing table as 'dont tell me about these guys'). but maybe we do. dunno. lots of implications.

Parameters:
replyTunnelId - tunnel to receive replies through
replyGateway - gateway for the reply tunnel
expiration - when the search should stop

buildMessage

protected DatabaseLookupMessage buildMessage(long expiration)
We're looking for a router, so lets build the lookup message (no need to tunnel route either, so just have replies sent back to us directly). This uses the similar overrides as the other buildMessage above.

Overrides:
buildMessage in class SearchJob

getBredth

protected int getBredth()
max # of concurrent searches

Overrides:
getBredth in class SearchJob

newPeersFound

protected void newPeersFound(int numNewPeers)
We've gotten a search reply that contained the specified number of peers that we didn't know about before.

Overrides:
newPeersFound in class SearchJob

getName

public java.lang.String getName()
Description copied from interface: Job
Descriptive name of the task

Specified by:
getName in interface Job
Overrides:
getName in class SearchJob

runJob

public void runJob()
Description copied from interface: Job
Actually perform the task. This call blocks until the Job is complete.


getState

protected SearchState getState()

getFacade

protected KademliaNetworkDatabaseFacade getFacade()

getExpiration

public long getExpiration()

getTimeoutMs

public long getTimeoutMs()

onlyQueryFloodfillPeers

static boolean onlyQueryFloodfillPeers(RouterContext ctx)

isCongested

static boolean isCongested(RouterContext ctx)

getPerPeerTimeoutMs

protected int getPerPeerTimeoutMs(Hash peer)

getPerPeerTimeoutMs

protected int getPerPeerTimeoutMs()
Let each peer take up to the average successful search RTT


searchNext

protected void searchNext()
Send the next search, or stop if its completed


continueSearch

protected void continueSearch()
Send a series of searches to the next available peers as selected by the routing table, but making sure no more than SEARCH_BREDTH are outstanding at any time


sendSearch

protected void sendSearch(RouterInfo router)
Send a search to the given peer


sendLeaseSearch

protected void sendLeaseSearch(RouterInfo router)
we're (probably) searching for a LeaseSet, so to be (overly) cautious, we're sending the request out through a tunnel w/ reply back through another tunnel.


sendRouterSearch

protected void sendRouterSearch(RouterInfo router)
we're searching for a router, so we can just send direct


buildMessage

protected DatabaseLookupMessage buildMessage(TunnelId replyTunnelId,
                                             Hash replyGateway,
                                             long expiration)
Build the database search message

Parameters:
replyTunnelId - tunnel to receive replies through
replyGateway - gateway for the reply tunnel
expiration - when the search should stop

replyFound

void replyFound(DatabaseSearchReplyMessage message,
                Hash peer)

fail

protected void fail()
Search totally failed


addDeferred

public int addDeferred(Job onFind,
                       Job onFail,
                       long expiration,
                       boolean isLease)

toString

public java.lang.String toString()
Overrides:
toString in class JobImpl

wasAttempted

boolean wasAttempted(Hash peer)

timeoutMs

long timeoutMs()

add

boolean add(Hash peer)

decrementOutstandingFloodfillSearches

void decrementOutstandingFloodfillSearches()