Adrien Béraud
4ef5a2216e
dht: revert to previous onNewNode behavior
2016-10-19 01:27:19 -04:00
Simon Désaulniers
09e41cc9f2
dht: appropriate time info in permanent put ops
2016-10-18 14:20:43 -04:00
Adrien Béraud
71ad92e286
dht: log cleanup
2016-10-17 17:55:22 -04:00
Adrien Béraud
d4f3d3aa98
dht: add missing iterator
2016-10-17 17:14:30 -04:00
Simon Désaulniers
c0612a1b2c
dht: use "refresh" operation for permanent put
2016-10-17 11:56:54 -04:00
Simon Désaulniers
6743cbbf50
dht: introduce "refresh" operation
...
This operation requests a node to refresh a value's time_point information of
its lifetime begining. This is intended to fix the *permanent put* operation
which has been having syncing issues since the introduction of an enhanced and
lighter put operation.
2016-10-17 11:56:54 -04:00
Adrien Béraud
da1bdf9d4c
dht: retry nodes more aggressively in case of failure
2016-10-16 22:36:52 -04:00
Adrien Béraud
96da4cd85d
build: make logs optional at build time
2016-10-16 22:36:42 -04:00
Adrien Béraud
cfb817d37f
dht: add buckets() utility
2016-10-16 22:36:30 -04:00
Adrien Béraud
9440c03048
dht: log cleanup
2016-10-16 22:36:24 -04:00
Adrien Béraud
7af856508e
dht: make connectivityChanged family dependent
2016-10-15 19:13:15 -04:00
Adrien Béraud
aaed64039a
dht: fix compilation on clang
2016-10-12 15:02:02 -04:00
Adrien Béraud
013b1dfe2e
Merge pull request #138 from sim590/connectivity-changes-detect
...
search: detect connectivity changes
2016-10-11 16:01:59 -04:00
Simon Désaulniers
733f699ff2
search: detect connectivity changes
...
The number of bad nodes is limited in order to help determine presence of
connectivity changes. See https://github.com/savoirfairelinux/opendht/issues/137
for details.
According to the tables, 25 is a good average value for big networks. If the
network is small, normal search expiration process will handle the situation.
2016-10-11 15:54:37 -04:00
Adrien Béraud
eed8ffc792
Merge pull request #139 from savoirfairelinux/recover-from-network-changes
...
dhtrunner: recover from network changes
2016-10-11 11:57:31 -04:00
Adrien Béraud
64118404fe
dhtrunner: recover from network changes
...
Using a list of bootstrap nodes, we recover from network loss by pinging those
nodes continuously until connected.
Written with Simon Désaulniers.
2016-10-10 01:51:26 -04:00
Simon Désaulniers
bb74dc846c
dht/dhtrunner: enable callback for pingNode
2016-10-10 00:23:53 -04:00
Adrien Béraud
659d893dee
network: transmit large values in multiple packets
2016-10-09 16:22:33 -04:00
Adrien Béraud
f53c7a6785
dht: keep track of storage usage in maintainStorage
2016-10-04 03:09:00 -04:00
Simon Désaulniers
dc0b650aa3
dht: make sure we are synced before sending put
2016-09-23 19:27:59 -04:00
Simon Désaulniers
87cf6faedd
dht/node: don't allow more than 3 auth errors
2016-09-23 19:27:59 -04:00
Simon Désaulniers
ca6aa6706d
dht/value: make 'find_node' queries satisfied by any queries
...
'find_node' requests don't ask for values. Then, the idea of a Query makes less
sense, but since we still need those to map status to network engine requests,
we introduce dummy queries which with the field Query::none. Those queries will
then only serve as key to network engine's requests.
2016-09-23 19:27:58 -04:00
Simon Désaulniers
16e55416a6
network_engine: fix bug token not sent for listen
2016-09-23 19:27:58 -04:00
Simon Désaulniers
3e24058175
dht: fix SearchNode's getListenTime computation
...
When no listen request have been sent yet, getListenTime should tell it's okay
to send pending listen request.
2016-09-23 19:27:55 -04:00
Simon Désaulniers
d1c2b20b28
dht: canGet must allow sending when no token
2016-09-23 13:48:04 -04:00
Adrien Béraud
23bb6e7a45
dht: use a map for storage
2016-09-16 01:29:33 -04:00
Adrien Béraud
6d38c18572
introduce SockAddr to store IP addresses
2016-09-15 01:38:26 -04:00
Simon Désaulniers
6f7ac21a85
dht: refactor code, terminating a Search (isDone)
...
More code encapsulated in objects method is better for keeping consistent
behavior.
2016-09-14 18:06:25 -04:00
Simon Désaulniers
7c5db84f33
dht: take multiple queries into account
...
The code preceding this would just not take multiple concurrent queries for a
same search into account. This is a bug fix for a bad behavior since the
queries.
2016-09-14 18:06:23 -04:00
Simon Désaulniers
4fecb2066b
dht: SearchNode::canGet criteria using 'update'
...
The criteria has been rewritten (again) exploiting the idea behind the parameter
'update' to take into account the case where we have new distinct requests to
send, then bypass some checks.
2016-09-14 18:06:12 -04:00
Adrien Béraud
eabdddcfb3
dht: fix warnings
2016-09-08 18:02:17 -04:00
Adrien Béraud
a9cc669a95
dht: clarify code
2016-09-08 15:52:18 -04:00
Simon Désaulniers
927c835b8f
dht: call all announce callback appropriately
2016-09-08 12:24:35 -04:00
Simon Désaulniers
5415eaaf60
dht: consider probe query in announce time
2016-09-08 12:24:35 -04:00
Simon Désaulniers
ec587f9b58
dht: make sure search can be flagged "done"
2016-09-08 12:24:35 -04:00
Simon Désaulniers
babfed73f3
dht: use SearchNode::acked for next announce time
2016-09-08 12:24:35 -04:00
Simon Désaulniers
2c0ce576e3
dht: enable data persistence
2016-09-08 12:24:35 -04:00
Simon Désaulniers
8f2cc937c5
dht: introduce light put
...
The 'put' operation will now first check if nodes already store the value we
have to announce. If so, we don't announce it to that node.
2016-09-08 12:24:35 -04:00
Simon Désaulniers
d60672e8ba
dht: remove unnecessary return value
2016-09-07 16:15:06 -04:00
Adrien Béraud
7919ca8de1
dht: fail pending operations before clearing them
2016-09-02 14:30:08 -04:00
Adrien Béraud
00ef660952
dht: hide more internal structures
2016-09-02 14:30:08 -04:00
kaldoran
2a3c301314
dht: Fix get and put at the same time on a same node
...
When the put is performed just after a get, then the get will not include the value which was in the put.
2016-09-01 18:02:23 -04:00
kaldoran
ae26a2b39a
Update of algorithm, by using new feature of api
2016-09-01 18:02:23 -04:00
Adrien Béraud
d1d0a67d35
dht: fix public address sorting
2016-09-01 12:46:39 -04:00
Simon Désaulniers
73686debaf
dht: before sending 'listen', send find_node
...
Since pagination, a bug was making the listen feature unusable running in
infinite loop. This also makes more sense than sending 'get' since the listen
request is paired with a Query defining the values that have to be fetched.
2016-08-30 15:51:50 -04:00
Adrien Béraud
b3401828e4
dht: set "connecting" state when waiting for a ping
2016-08-25 00:42:50 -04:00
Simon Désaulniers
8d7deffa6a
dht: properly find pending satisfying query
...
In SearchNode::canGet, we don't send a 'get' request to a search node if a
satisfying query for the query we want to send is already pending. This patch
honors this behavior.
2016-08-22 18:45:29 -04:00
Simon Désaulniers
669c08210e
dht: erase find node request status when done
...
This was memory leak and produced inconsistency in SearchNode::canGet timings.
2016-08-22 18:26:53 -04:00
Adrien Béraud
690682408b
build with -pedantic-errors, fix pedantic build errors
2016-08-16 10:26:44 -04:00
Simon Désaulniers
e9bf042ac5
dht: document SearchNode::canGet
2016-08-15 13:44:52 -04:00
Adrien Béraud
Simon Désaulniers
kaldoran