Victor Shyba
|
d8c1aaebc2
|
routing table: mark private methods
|
2022-08-11 20:38:42 -03:00 |
|
Victor Shyba
|
d7b65c15d2
|
return none instead of raising
|
2022-08-11 20:38:42 -03:00 |
|
Victor Shyba
|
972db80246
|
move add peer logic to routing table
|
2022-08-11 20:38:42 -03:00 |
|
Victor Shyba
|
0d343ecb2f
|
simplify iterative find constructor
|
2022-08-11 20:38:42 -03:00 |
|
Jonathan Moody
|
e5e9873f79
|
Simplify by eliminating AsyncGenerator base and generator function. Remove any new places enforcing max_results.
|
2022-05-20 17:23:39 -04:00 |
|
Jonathan Moody
|
fe07aac79c
|
Define and use lbry.utils.aclosing() in lieu of official contextlib.aclosing().
|
2022-05-20 17:23:39 -04:00 |
|
Jonathan Moody
|
91a6eae831
|
Fix lint issue in iterative_find.py.
|
2022-05-20 17:23:39 -04:00 |
|
Jonathan Moody
|
5852fcd287
|
Don't wait on running_tasks after cancel(). Sometimes a CancelledError exception is received, which is unhelpful, and complicates shutting down the generator.
|
2022-05-20 17:23:39 -04:00 |
|
Jonathan Moody
|
4767bb9dee
|
Wrap "async for" over IterativeXXXFinder in try/finally ensuring aclose().
|
2022-05-20 17:23:39 -04:00 |
|
Jonathan Moody
|
82d7f81f41
|
Correct call to _aclose() in response to TransportNotConnected.
|
2022-05-20 17:23:39 -04:00 |
|
Jonathan Moody
|
b036961954
|
Tighten up IterativeFinder logic to respect max_records better, and wait after task cancel().
Also make IterativeFinder a proper AsyncGenerator. This gives it an offically recognized aclose() method and could help with clean finalization.
|
2022-05-20 17:23:39 -04:00 |
|
Victor Shyba
|
f05943ff79
|
implement announcer as a consumer task on gather
|
2022-03-02 13:00:34 -03:00 |
|
Victor Shyba
|
7ded8a1333
|
make active an explicit ordered dict
|
2022-03-02 13:00:34 -03:00 |
|
Victor Shyba
|
c2478d4add
|
remove unused search rounds
|
2022-03-02 13:00:34 -03:00 |
|
Victor Shyba
|
8b1009161a
|
better representation of kademliapeer on debug logs
|
2022-03-02 13:00:34 -03:00 |
|
Victor Shyba
|
868a620e91
|
add a way to wait announcements to finish so tests are reliable
|
2022-03-02 13:00:34 -03:00 |
|
Victor Shyba
|
612dbcb2f3
|
allow running some extra probes for k replacements
|
2022-03-02 13:00:34 -03:00 |
|
Victor Shyba
|
b3614d965d
|
remove all references to bottoming out
|
2022-03-02 13:00:34 -03:00 |
|
Victor Shyba
|
5d7137255e
|
no stop condition, let it exhaust
|
2022-03-02 13:00:34 -03:00 |
|
Victor Shyba
|
6ff867ef55
|
bottoming out is now warning and no results for peer search
|
2022-03-02 13:00:34 -03:00 |
|
Victor Shyba
|
c14915df29
|
don't probe peers too far from the top closest
|
2022-03-02 13:00:34 -03:00 |
|
Victor Shyba
|
7d4966e2ae
|
use a dict for the active queue
|
2022-03-02 13:00:34 -03:00 |
|
Victor Shyba
|
3876e0317d
|
log bottom out of peer search in debug, show short key id for find value
|
2022-03-02 13:00:34 -03:00 |
|
Victor Shyba
|
0b2b10f759
|
bump bottom out limit of peer search so people can use 100 concurrent announcers
|
2022-03-02 13:00:34 -03:00 |
|
Victor Shyba
|
9a79b33664
|
wait until k peers are ready. do not double add peers
|
2022-03-02 13:00:34 -03:00 |
|
Victor Shyba
|
af1a6edd15
|
only return good (contacted) peers
|
2022-03-02 13:00:34 -03:00 |
|
Victor Shyba
|
b78929f4d5
|
reset closest peer on failure
|
2022-03-02 13:00:34 -03:00 |
|
Victor Shyba
|
fb6e342043
|
add peers from shortlist regardless, but check from other nodes
|
2022-03-02 13:00:34 -03:00 |
|
Victor Shyba
|
511e57c231
|
fix distance sorting and improve logging
|
2022-03-02 13:00:34 -03:00 |
|
Victor Shyba
|
d762d675c4
|
closest peer is only ready when it was contacted and isn't known to be bad
|
2022-03-02 13:00:34 -03:00 |
|
Victor Shyba
|
3fdadee87c
|
dont probe and ignore bad peers
|
2022-03-02 13:00:34 -03:00 |
|
Victor Shyba
|
1aa4d9d585
|
simplify, genaralize to any size and fix tests
|
2022-02-28 13:06:51 -03:00 |
|
Victor Shyba
|
ca65c1ebc5
|
replace duplicated code
|
2022-02-28 13:06:51 -03:00 |
|
Victor Shyba
|
c0f512ace7
|
bump DHT peer manager cache to 16384
|
2022-02-02 16:54:42 -03:00 |
|
Victor Shyba
|
c9d637b4da
|
add gauge for queue size
|
2022-02-02 11:56:42 -03:00 |
|
Victor Shyba
|
ae3e8fadf5
|
count announcements and how many peers we were able to announce to
|
2022-02-02 11:56:42 -03:00 |
|
Victor Shyba
|
0618053bd4
|
remove request_flight metric
|
2022-01-12 12:41:04 -03:00 |
|
Victor Shyba
|
fd9dcbf9a8
|
add granular metric for stored blob prefix, for network announcements calculation
|
2022-01-12 12:39:23 -03:00 |
|
Victor Shyba
|
beb8583436
|
change colliding bits metric to gauge
|
2022-01-12 12:39:23 -03:00 |
|
Victor Shyba
|
b44e2c0b38
|
count bit collisions between 8 and 16
|
2022-01-12 12:39:23 -03:00 |
|
Victor Shyba
|
06e94640b5
|
add counter for peers with colliding bytes
|
2022-01-12 12:39:23 -03:00 |
|
Victor Shyba
|
ff36bdc802
|
add requests in flight and error
|
2022-01-12 12:39:23 -03:00 |
|
Victor Shyba
|
46f576de46
|
add request received
|
2022-01-12 12:39:23 -03:00 |
|
Victor Shyba
|
7b09c34fce
|
add request_sent and request_time metric on dht
|
2022-01-12 12:39:23 -03:00 |
|
Victor Shyba
|
a22f50aa84
|
add storing_peers and peer_manager_keys
|
2022-01-12 12:39:23 -03:00 |
|
Victor Shyba
|
2d9130b4e0
|
prometheus: move blobs_stored and peers to SDK. add buckets_in_routing_table
|
2022-01-12 12:39:23 -03:00 |
|
Victor Shyba
|
a8523996a9
|
extract cache values, increase peer cache to 2048
|
2022-01-07 12:58:52 -03:00 |
|
Victor Shyba
|
f586de2bbe
|
DHT bugfix: failures tracking should be bound to 2048 LRU cache size
|
2022-01-07 12:46:00 -03:00 |
|
Victor Shyba
|
54d6fb9da4
|
do not limit DHT results by K, respect max_results
|
2021-12-09 14:34:55 -03:00 |
|
Victor Shyba
|
3d5c9cc1c2
|
clarify DHT debug logging on key and operation
|
2021-12-09 14:32:30 -03:00 |
|