Compare commits

...
31 Commits
Author SHA1 Message Date
Alex Grin de12e25450 v0.38.6 2019-07-29 15:11:18 -04:00
Oleg Silkin 2545ebbece Adds unittests & fixes comment_client functions 2019-07-29 14:55:42 -04:00
Oleg Silkin 14ec6ce43b Adds unit-testing for delete comment signatures 2019-07-29 14:55:42 -04:00
Oleg Silkin 0cac1e65aa Imports comment_client and uses dot notation for rest of modules 2019-07-29 14:55:42 -04:00
Oleg Silkin 1e4d57ab83 Simplifies signing function into one to make API symmetric 2019-07-29 14:55:42 -04:00
Oleg Silkin 49b42b7e03 Simplifies 2019-07-29 14:55:42 -04:00
Oleg Silkin 89c0b5dbdf adds comment abandon function 2019-07-29 14:55:42 -04:00
Lex Berezhny 8138675808 added index for claim_type searches 2019-07-29 14:55:07 -04:00
Jack Robison 726129836a Merge pull request #2323 from lbryio/refactor_ttfb
refactor and fix ttfb for new api
2019-07-28 22:28:02 -04:00
Jack Robison 356b3c85e5 use app default order_by 2019-07-28 21:42:04 -04:00
Jack Robison 1a0680ead9 update claim_search_performance.py, add new test cases shown by time_to_first_byte.py 2019-07-28 21:22:07 -04:00
Jack Robison c876d891fa use perf_counter 2019-07-28 21:18:29 -04:00
Jack Robison a08232f164 fix claim search kwargs in time to first byte script 2019-07-28 21:18:13 -04:00
Victor Shyba 0f8cf46626 refactor and fix ttfb for new api 2019-07-28 17:33:37 -04:00
Alex Grin 95017b5fc8 v0.38.5 2019-07-25 13:22:54 -04:00
Jack Robison 1374d337d8 Merge pull request #2348 from lbryio/probemon
Probe if peers can be connected to during blob downloads to more quickly skip bad peers
2019-07-25 13:21:42 -04:00
Jack Robison 35e7cc685f fix logging error 2019-07-24 22:12:51 -04:00
Victor Shyba 2ac4883bda cleanup and fix prober 2019-07-24 22:32:20 -03:00
Victor Shyba 5a7212aebb gotta probe 'em all 2019-07-24 20:26:38 -03:00
Lex Berezhny 3f39cc2955 v0.38.4 2019-07-24 18:02:03 -04:00
Lex Berezhny 8a5bfe884a save channel keys into wallet before broadcasting 2019-07-24 18:00:06 -04:00
Lex Berezhny 046dd53913 add index to filter by fee_amount and order by release_time 2019-07-24 14:27:59 -04:00
Alex Grintsvayg d2f174635c add script to monitor wallet servers for slow queries 2019-07-24 13:36:22 -04:00
Lex Berezhny 01a760d8f0 v0.38.3 2019-07-23 17:44:38 -04:00
Victor Shyba 9215d2c9f9 log that checkpoint succeeded 2019-07-23 17:44:13 -04:00
Victor Shyba 0bb09ae91e use checkpoint to speed up 2019-07-23 17:44:13 -04:00
Victor Shyba afeff61268 fix repair and tests 2019-07-23 17:44:13 -04:00
Victor Shyba 6f5a595975 check from final size instead of partials 2019-07-23 17:44:13 -04:00
Victor Shyba 8a69ec062e download headers to disk and repair instead of connecting each 2019-07-23 17:44:13 -04:00
Victor Shyba 45fb673e49 dont fail when s3 is out of reach 2019-07-23 17:44:13 -04:00
Lex Berezhny 93b5f1809e revert to v0.38.2 so we can release v0.38.3 again 2019-07-23 17:43:04 -04:00
15 changed files with 489 additions and 246 deletions

No files matched your search

+1 -1
View File
@@ -1,4 +1,4 @@
__name__ = "lbry"
__version__ = "0.38.3"
__version__ = "0.38.6"
version = tuple(__version__.split('.'))
+5 -3
View File
@@ -187,7 +187,7 @@ class BlobExchangeClientProtocol(asyncio.Protocol):
return await self._download_blob()
except OSError as e:
# i'm not sure how to fix this race condition - jack
log.warning("race happened downloading %s from %s:%i", blob_hash, self.peer_address, self.peer_port)
log.warning("race happened downloading %s from %s:%s", blob_hash, self.peer_address, self.peer_port)
# return self._blob_bytes_received, self.transport
raise
except asyncio.TimeoutError:
@@ -220,7 +220,7 @@ class BlobExchangeClientProtocol(asyncio.Protocol):
@cache_concurrent
async def request_blob(loop: asyncio.BaseEventLoop, blob: 'AbstractBlob', address: str, tcp_port: int,
async def request_blob(loop: asyncio.BaseEventLoop, blob: typing.Optional['AbstractBlob'], address: str, tcp_port: int,
peer_connect_timeout: float, blob_download_timeout: float,
connected_transport: asyncio.Transport = None, connection_id: int = 0,
connection_manager: typing.Optional['ConnectionManager'] = None)\
@@ -242,7 +242,9 @@ async def request_blob(loop: asyncio.BaseEventLoop, blob: 'AbstractBlob', addres
if not connected_transport:
await asyncio.wait_for(loop.create_connection(lambda: protocol, address, tcp_port),
peer_connect_timeout, loop=loop)
if blob.get_is_verified() or not blob.is_writeable():
connected_transport = protocol.transport
if blob is None or blob.get_is_verified() or not blob.is_writeable():
# blob is None happens when we are just opening a connection
# file exists but not verified means someone is writing right now, give it time, come back later
return 0, connected_transport
return await protocol.download_blob(blob)
+13 -9
View File
@@ -30,17 +30,19 @@ class BlobDownloader:
self.is_running = asyncio.Event(loop=self.loop)
def should_race_continue(self, blob: 'AbstractBlob'):
if len(self.active_connections) >= self.config.max_connections_per_download:
max_probes = self.config.max_connections_per_download * (1 if self.connections else 10)
if len(self.active_connections) >= max_probes:
return False
return not (blob.get_is_verified() or not blob.is_writeable())
async def request_blob_from_peer(self, blob: 'AbstractBlob', peer: 'KademliaPeer', connection_id: int = 0):
async def request_blob_from_peer(self, blob: 'AbstractBlob', peer: 'KademliaPeer', connection_id: int = 0,
just_probe: bool = False):
if blob.get_is_verified():
return
transport = self.connections.get(peer)
start = self.loop.time()
bytes_received, transport = await request_blob(
self.loop, blob, peer.address, peer.tcp_port, self.config.peer_connect_timeout,
self.loop, blob if not just_probe else None, peer.address, peer.tcp_port, self.config.peer_connect_timeout,
self.config.blob_download_timeout, connected_transport=transport, connection_id=connection_id,
connection_manager=self.blob_manager.connection_manager
@@ -91,16 +93,18 @@ class BlobDownloader:
if batch:
self.peer_queue.put_nowait(list(batch))
log.debug(
"running, %d peers, %d ignored, %d active",
len(batch), len(self.ignored), len(self.active_connections)
"running, %d peers, %d ignored, %d active, %s connections",
len(batch), len(self.ignored), len(self.active_connections), len(self.connections)
)
for peer in sorted(batch, key=lambda peer: self.scores.get(peer, 0), reverse=True):
if peer in self.ignored or peer in self.active_connections:
continue
if not self.should_race_continue(blob):
break
if peer not in self.active_connections and peer not in self.ignored:
log.debug("request %s from %s:%i", blob_hash[:8], peer.address, peer.tcp_port)
t = self.loop.create_task(self.request_blob_from_peer(blob, peer, connection_id))
self.active_connections[peer] = t
log.debug("request %s from %s:%i", blob_hash[:8], peer.address, peer.tcp_port)
just_probe = len(self.connections) == 0
t = self.loop.create_task(self.request_blob_from_peer(blob, peer, connection_id, just_probe))
self.active_connections[peer] = t
await self.new_peer_or_finished()
self.cleanup_active()
log.debug("downloaded %s", blob_hash[:8])
+50 -40
View File
@@ -1,3 +1,4 @@
import hashlib
import os
import asyncio
import logging
@@ -96,12 +97,10 @@ class DatabaseComponent(Component):
self.storage = None
HEADERS_URL = "https://headers.lbry.io/blockchain_headers_latest"
HEADER_SIZE = 112
class HeadersComponent(Component):
component_name = HEADERS_COMPONENT
HEADERS_URL = "https://headers.lbry.io/blockchain_headers_latest"
CHECKPOINT = ('100b33ca3d0b86a48f0d6d6f30458a130ecb89d5affefe4afccb134d5a40f4c2', 600_000)
def __init__(self, component_manager):
super().__init__(component_manager)
@@ -135,31 +134,26 @@ class HeadersComponent(Component):
} if progress is not None and progress < 100 else {}
async def fetch_headers_from_s3(self):
local_header_size = self.headers.bytes_size
local_header_size = self.local_header_file_size()
resume_header = {"Range": f"bytes={local_header_size}-"}
async with utils.aiohttp_request('get', HEADERS_URL, headers=resume_header) as response:
if response.status == 406 or response.content_length < HEADER_SIZE: # our file is bigger
async with utils.aiohttp_request('get', self.HEADERS_URL, headers=resume_header) as response:
if response.status == 406 or response.content_length < self.headers.header_size: # our file is bigger
log.warning("s3 is more out of date than we are")
return
if response.content_length % HEADER_SIZE != 0:
final_size_after_download = response.content_length + local_header_size
if final_size_after_download % self.headers.header_size != 0:
log.warning("s3 appears to have corrupted header")
return
final_size_after_download = response.content_length + local_header_size
write_mode = "wb"
if local_header_size > 0:
log.info("Resuming download of %i bytes from s3", response.content_length)
buffer, header_size = b'', self.headers.header_size
async for chunk in response.content.iter_any():
chunk = buffer + chunk
remaining = len(chunk) % header_size
chunk, buffer = chunk[:-remaining], bytes(chunk[-remaining:])
if not chunk:
continue
if not await self.headers.connect(len(self.headers), chunk):
log.warning("Error connecting downloaded headers from at %s.", self.headers.height)
return
self._headers_progress_percent = self._round_progress(
self.headers.bytes_size, final_size_after_download
)
write_mode = "a+b"
with open(self.headers_file, write_mode) as fd:
while not response.content.at_eof():
local_header_size += fd.write(await response.content.readany())
self._headers_progress_percent = self._round_progress(
local_header_size, final_size_after_download
)
def local_header_file_size(self) -> int:
if os.path.isfile(self.headers_file):
@@ -167,13 +161,11 @@ class HeadersComponent(Component):
return 0
async def get_downloadable_header_height(self) -> typing.Optional[int]:
try:
async with utils.aiohttp_request('HEAD', HEADERS_URL) as response:
if response.status != 200:
log.warning("Header download error, unexpected response code: %s", response.status)
return response.content_length // HEADER_SIZE
except OSError:
log.exception("Failed to download headers using https.")
async with utils.aiohttp_request('HEAD', self.HEADERS_URL) as response:
if response.status != 200:
log.warning("Header download error, unexpected response code: %s", response.status)
return -1
return response.content_length // self.headers.header_size
async def should_download_headers_from_s3(self) -> bool:
if self.conf.blockchain_name != "lbrycrd_main":
@@ -182,7 +174,7 @@ class HeadersComponent(Component):
if not s3_headers_depth:
return False
local_height = self.headers.height
local_height = self.local_header_file_size() // self.headers.header_size
remote_height = await self.get_downloadable_header_height()
if remote_height is not None:
log.info("remote height: %i, local height: %i", remote_height, local_height)
@@ -190,24 +182,42 @@ class HeadersComponent(Component):
return True
return False
def verify_checkpoint(self):
expected_hash, at_height = self.CHECKPOINT
if self.local_header_file_size() // self.headers.header_size < at_height:
return False
hash = hashlib.sha256()
chunk_size = self.headers.header_size * 1000
with open(self.headers_file, 'rb') as header_file:
data = header_file.read(chunk_size)
while data and header_file.tell() <= at_height * self.headers.header_size:
hash.update(data)
data = header_file.read(chunk_size)
return hash.hexdigest() == expected_hash
async def start(self):
if not os.path.exists(self.headers_dir):
os.mkdir(self.headers_dir)
if os.path.exists(self.old_file):
log.warning("Moving old headers from %s to %s.", self.old_file, self.headers_file)
os.rename(self.old_file, self.headers_file)
await self.headers.open()
await self.headers.repair()
if await self.should_download_headers_from_s3():
try:
try:
if await self.should_download_headers_from_s3():
self.is_downloading_headers = True
await self.fetch_headers_from_s3()
except Exception as err:
log.error("failed to fetch headers from s3: %s", err)
finally:
self.is_downloading_headers = False
await self.headers.close()
except Exception as err:
log.error("failed to fetch headers from s3: %s", err)
finally:
self.is_downloading_headers = False
# fixme: workaround, this should happen before download but happens after because headers.connect fail
if not self.verify_checkpoint():
log.info("Checkpoint failed, verifying headers using slower method.")
await self.headers.open()
await self.headers.repair()
await self.headers.close()
else:
log.info("Header checkpoint verified.")
async def stop(self):
pass
+39 -8
View File
@@ -33,7 +33,7 @@ from lbry.extras.daemon.Components import EXCHANGE_RATE_MANAGER_COMPONENT, UPNP_
from lbry.extras.daemon.ComponentManager import RequiredCondition
from lbry.extras.daemon.ComponentManager import ComponentManager
from lbry.extras.daemon.json_response_encoder import JSONResponseEncoder
from lbry.extras.daemon.comment_client import jsonrpc_post, sign_comment, is_comment_signed_by_channel
from lbry.extras.daemon import comment_client
from lbry.extras.daemon.undecorated import undecorated
from lbry.wallet.transaction import Transaction, Output, Input
from lbry.wallet.account import Account as LBCAccount
@@ -1946,9 +1946,9 @@ class Daemon(metaclass=JSONRPCServerType):
if not preview:
await tx.sign([account])
await self.broadcast_or_release(account, tx, blocking)
account.add_channel_private_key(txo.private_key)
self.default_wallet.save()
await self.broadcast_or_release(account, tx, blocking)
await self.storage.save_claims([self._old_get_temp_claim_info(
tx, txo, claim_address, claim, name, dewies_to_lbc(amount)
)])
@@ -2085,9 +2085,9 @@ class Daemon(metaclass=JSONRPCServerType):
if not preview:
await tx.sign([account])
await self.broadcast_or_release(account, tx, blocking)
account.add_channel_private_key(new_txo.private_key)
self.default_wallet.save()
await self.broadcast_or_release(account, tx, blocking)
await self.storage.save_claims([self._old_get_temp_claim_info(
tx, new_txo, claim_address, new_txo.claim, new_txo.claim_name, dewies_to_lbc(amount)
)])
@@ -3416,7 +3416,7 @@ class Daemon(metaclass=JSONRPCServerType):
]
}
"""
result = await jsonrpc_post(
result = await comment_client.jsonrpc_post(
self.conf.comment_server,
"get_claim_comments",
claim_id=claim_id,
@@ -3432,7 +3432,7 @@ class Daemon(metaclass=JSONRPCServerType):
continue
resolve_response = await self.resolve([channel_url])
if isinstance(resolve_response[channel_url], Output):
comment['is_channel_signature_valid'] = is_comment_signed_by_channel(
comment['is_channel_signature_valid'] = comment_client.is_comment_signed_by_channel(
comment, resolve_response[channel_url]
)
else:
@@ -3490,12 +3490,43 @@ class Daemon(metaclass=JSONRPCServerType):
'channel_id': channel.claim_id,
'channel_name': channel.claim_name,
})
sign_comment(comment_body, channel)
response = await jsonrpc_post(self.conf.comment_server, 'create_comment', **comment_body)
comment_client.sign_comment(comment_body, channel)
response = await comment_client.jsonrpc_post(self.conf.comment_server, 'create_comment', comment_body)
if 'signature' in response:
response['is_claim_signature_valid'] = is_comment_signed_by_channel(response, channel)
response['is_claim_signature_valid'] = comment_client.is_comment_signed_by_channel(response, channel)
return response
async def jsonrpc_comment_abandon(self, comment_id):
""""
Delete a comment published under your channel identity
Usage:
comment_delete (<comment_id> | --comment_id=<comment_id>)
Options:
--comment_id=<comment_id> : (str) The ID of the comment to be deleted.
Returns:
(dict) Object with the `comment_id` passed in as the key, and a flag indicating if it was deleted
{
<comment_id> (str): {
"deleted": (bool)
}
}
"""
abandon_comment_body = {'comment_id': comment_id}
channel = await comment_client.jsonrpc_post(
self.conf.comment_server, 'get_channel_from_comment_id', comment_id=comment_id
)
if 'error' in channel:
return {comment_id: {'deleted': False}}
channel = await self.get_channel_or_none(None, **channel)
abandon_comment_body.update({
'channel_id': channel.claim_id,
'channel_name': channel.claim_name,
})
comment_client.sign_comment(abandon_comment_body, channel, abandon=True)
return await comment_client.jsonrpc_post(self.conf.comment_server, 'delete_comment', abandon_comment_body)
async def broadcast_or_release(self, account, tx, blocking=False):
try:
+20 -10
View File
@@ -18,13 +18,18 @@ def get_encoded_signature(signature):
return ecdsa.util.sigencode_der(r, s, len(signature) * 4)
def is_comment_signed_by_channel(comment: dict, channel: Output):
def cid2hash(claim_id: str) -> bytes:
return binascii.unhexlify(claim_id.encode())[::-1]
def is_comment_signed_by_channel(comment: dict, channel: Output, abandon=False):
if type(channel) is Output:
try:
signing_field = comment['comment_id'] if abandon else comment['comment']
pieces = [
comment['signing_ts'].encode(),
channel.claim_hash,
comment['comment'].encode()
cid2hash(comment['channel_id']),
signing_field.encode()
]
return Output.is_signature_valid(
get_encoded_signature(comment['signature']),
@@ -36,16 +41,21 @@ def is_comment_signed_by_channel(comment: dict, channel: Output):
return False
def sign_comment(comment: dict, channel: Output):
timestamp = str(int(time.time())).encode()
pieces = [timestamp, channel.claim_hash, comment['comment'].encode()]
def sign_comment(comment: dict, channel: Output, abandon=False):
timestamp = str(int(time.time()))
signing_field = comment['comment_id'] if abandon else comment['comment']
pieces = [timestamp.encode(), channel.claim_hash, signing_field.encode()]
digest = sha256(b''.join(pieces))
signature = channel.private_key.sign_digest_deterministic(digest, hashfunc=hashlib.sha256)
comment['signature'] = binascii.hexlify(signature).decode()
comment['signing_ts'] = timestamp.decode()
comment.update({
'signature': binascii.hexlify(signature).decode(),
'signing_ts': timestamp
})
async def jsonrpc_post(url: str, method: str, **params) -> any:
async def jsonrpc_post(url: str, method: str, params: dict = None, **kwargs) -> any:
params = params or {}
params.update(kwargs)
json_body = {'jsonrpc': '2.0', 'id': None, 'method': method, 'params': params}
headers = {'Content-Type': 'application/json'}
async with utils.aiohttp_request('POST', url, json=json_body, headers=headers) as response:
@@ -53,5 +63,5 @@ async def jsonrpc_post(url: str, method: str, **params) -> any:
result = await response.json()
return result['result'] if 'result' in result else result
except Exception as cte:
log.exception('Unable to decode respose from server: %s', cte)
log.exception('Unable to decode response from server: %s', cte)
return await response.text()
+3 -3
View File
@@ -121,6 +121,9 @@ class SQLDB:
create unique index if not exists claim_effective_amount_idx on claim (effective_amount, claim_hash, release_time);
create unique index if not exists claim_release_time_idx on claim (release_time, claim_hash);
create unique index if not exists claim_trending_global_mixed_idx on claim (trending_global, trending_mixed, claim_hash);
create unique index if not exists filter_fee_amount_order_release_time_idx on claim (fee_amount, release_time, claim_hash);
create unique index if not exists claim_type_trending_idx on claim (claim_type, trending_global, trending_mixed, claim_hash);
-- TODO: verify that all indexes below are used
create index if not exists claim_height_normalized_idx on claim (height, normalized asc);
@@ -131,11 +134,8 @@ class SQLDB:
create index if not exists claim_timestamp_idx on claim (timestamp);
create index if not exists claim_public_key_hash_idx on claim (public_key_hash);
create index if not exists claim_claim_type_idx on claim (claim_type);
create index if not exists claim_stream_type_idx on claim (stream_type);
create index if not exists claim_media_type_idx on claim (media_type);
create index if not exists claim_fee_amount_idx on claim (fee_amount);
create index if not exists claim_fee_currency_idx on claim (fee_currency);
create index if not exists claim_signature_valid_idx on claim (signature_valid);
"""
+77 -46
View File
@@ -1,5 +1,6 @@
import os
import time
import textwrap
import argparse
import asyncio
import logging
@@ -9,7 +10,7 @@ from lbry.wallet.ledger import MainNetLedger
log = logging.getLogger(__name__)
log.addHandler(logging.StreamHandler())
log.setLevel(logging.INFO)
log.setLevel(logging.CRITICAL)
DEFAULT_ANY_TAGS = [
'blockchain',
@@ -68,43 +69,46 @@ def get_args(limit=20):
args = []
any_tags_combinations = [DEFAULT_ANY_TAGS, COMMON_AND_RARE, RARE_ANY_TAGS, COMMON_AND_RARE2, CITY_FIX, []]
not_tags_combinations = [MATURE_TAGS, []]
for no_totals in [True]:
for offset in [0, 100]:
for any_tags in any_tags_combinations:
for not_tags in not_tags_combinations:
for order_by in ORDER_BY:
kw = {
'order_by': order_by,
'offset': offset,
'limit': limit,
'no_totals': no_totals
}
if not_tags:
kw['not_tags'] = not_tags
if any_tags:
kw['any_tags'] = any_tags
args.append(kw)
print(len(args), "argument combinations")
for no_fee in [False, True]:
for claim_type in [None, 'stream', 'channel']:
for no_totals in [True]:
for offset in [0, 100]:
for any_tags in any_tags_combinations:
for not_tags in not_tags_combinations:
for order_by in ORDER_BY:
kw = {
'order_by': order_by,
'offset': offset,
'limit': limit,
'no_totals': no_totals
}
if not_tags:
kw['not_tags'] = not_tags
if any_tags:
kw['any_tags'] = any_tags
if claim_type:
kw['claim_type'] = claim_type
if no_fee:
kw['fee_amount'] = 0
args.append(kw)
print(f"-- Trying {len(args)} argument combinations")
return args
def _search(kwargs):
start = time.time()
start = time.perf_counter()
error = None
try:
search_to_bytes(kwargs)
t = time.time() - start
return t, kwargs
except Exception as err:
return -1, f"failed: error={str(type(err))}({str(err)})"
error = str(err)
return time.perf_counter() - start, kwargs, error
async def search(executor, kwargs):
try:
return await asyncio.get_running_loop().run_in_executor(
executor, _search, kwargs
)
except Exception as err:
return f"failed (err={str(type(err))}({err}))- {kwargs}"
return await asyncio.get_running_loop().run_in_executor(
executor, _search, kwargs
)
async def main(db_path, max_query_time):
@@ -115,23 +119,50 @@ async def main(db_path, max_query_time):
tasks = [search(query_executor, constraints) for constraints in get_args()]
try:
results = await asyncio.gather(*tasks)
for ts, constraints in results:
if ts >= max_query_time:
sql = interpolate(*_get_claims("""
claimtrie.claim_hash as is_controlling,
claimtrie.last_take_over_height,
claim.claim_hash, claim.txo_hash,
claim.claims_in_channel,
claim.height, claim.creation_height,
claim.activation_height, claim.expiration_height,
claim.effective_amount, claim.support_amount,
claim.trending_group, claim.trending_mixed,
claim.trending_local, claim.trending_global,
claim.short_url, claim.canonical_url,
claim.channel_hash, channel.txo_hash AS channel_txo_hash,
channel.height AS channel_height, claim.signature_valid
""", **constraints))
print(f"Query took {int(ts * 1000)}ms\n{sql}")
query_times = [
{
'sql': interpolate(*_get_claims("""
claimtrie.claim_hash as is_controlling,
claimtrie.last_take_over_height,
claim.claim_hash, claim.txo_hash,
claim.claims_in_channel,
claim.height, claim.creation_height,
claim.activation_height, claim.expiration_height,
claim.effective_amount, claim.support_amount,
claim.trending_group, claim.trending_mixed,
claim.trending_local, claim.trending_global,
claim.short_url, claim.canonical_url,
claim.channel_hash, channel.txo_hash AS channel_txo_hash,
channel.height AS channel_height, claim.signature_valid
""", **constraints)),
'duration': ts,
'error': error
}
for ts, constraints, error in results
]
errored = [query_info for query_info in query_times if query_info['error']]
errors = {str(query_info['error']): [] for query_info in errored}
for error in errored:
errors[str(error['error'])].append(error['sql'])
slow = [
query_info for query_info in query_times
if not query_info['error'] and query_info['duration'] > (max_query_time / 2.0)
]
fast = [
query_info for query_info in query_times
if not query_info['error'] and query_info['duration'] <= (max_query_time / 2.0)
]
print(f"-- {len(fast)} queries were fast")
slow.sort(key=lambda query_info: query_info['duration'], reverse=True)
print(f"-- Failing queries:")
for error in errors:
print(f"-- Failure: \"{error}\"")
for failing_query in errors[error]:
print(f"{textwrap.dedent(failing_query)};\n")
print()
print(f"-- Slow queries:")
for slow_query in slow:
print(f"-- Query took {slow_query['duration']}\n{textwrap.dedent(slow_query['sql'])};\n")
finally:
query_executor.shutdown()
@@ -139,7 +170,7 @@ async def main(db_path, max_query_time):
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument('--db_path', dest='db_path', default=os.path.expanduser('~/claims.db'), type=str)
parser.add_argument('--max_time', dest='max_time', default=0.0, type=float)
parser.add_argument('--max_time', dest='max_time', default=0.25, type=float)
args = parser.parse_args()
db_path = args.db_path
max_query_time = args.max_time
+49
View File
@@ -0,0 +1,49 @@
import os, asyncio, aiohttp, json, slack, sqlparse
async def listen(slack_client, url):
async with aiohttp.ClientSession(timeout=aiohttp.ClientTimeout(3)) as session:
print(f"connecting to {url}")
try:
ws = await session.ws_connect(url)
except (aiohttp.ClientConnectorError, asyncio.TimeoutError):
print(f"failed to connect to {url}")
return
print(f"connected to {url}")
async for msg in ws:
r = json.loads(msg.data)
try:
queries = r["api"]["search"]["interrupted_queries"]
except KeyError:
continue
for q in queries:
# clean = re.sub(r"\s+", " ", q)
clean = sqlparse.format(q, reindent=True, keyword_case='upper')
print(f'{url}: {clean}')
response = await slack_client.chat_postMessage(
username=url,
icon_emoji=":hourglass_flowing_sand:",
channel='#clubhouse-de-obscure',
text="*Query timed out:* " + clean
)
if not response["ok"]:
print("SLACK ERROR:\n", response)
print()
async def main():
try:
slack_client = slack.WebClient(token=os.environ['SLACK_TOKEN'], run_async=True)
except KeyError:
print("Error: SLACK_TOKEN env var required")
return
num_servers = 5
tasks = []
for i in range(1, num_servers+1):
tasks.append(asyncio.create_task(listen(slack_client, f'http://spv{i}.lbry.com:50005')))
await asyncio.gather(*tasks)
asyncio.run(main())
+86 -103
View File
@@ -1,43 +1,19 @@
import os
import sys
import json
import argparse
import asyncio
import aiohttp
import time
import aiohttp
from aiohttp import ClientConnectorError
from lbry import __version__
from lbry.blob.blob_file import MAX_BLOB_SIZE
from lbry.conf import Config
from lbry.schema.uri import parse_lbry_uri
from lbry.extras.daemon.client import daemon_rpc
from lbry.extras import system_info
def extract_uris(response):
uris = list()
for key in response:
for value in response[key]:
uris.append(value)
return uris
async def get_frontpage_uris():
session = aiohttp.ClientSession()
try:
response = await session.get("https://api.lbry.com/file/list_homepage", timeout=10.0)
if response.status != 200:
print("API returned non 200 code!!")
return
body = await response.json()
await session.close()
uris = extract_uris(body['data']['Uris'])
return uris
finally:
await session.close()
async def report_to_slack(output, webhook):
payload = {
"text": f"lbrynet {__version__} ({system_info.get_platform()['platform']}) time to first byte:\n{output}"
@@ -58,120 +34,127 @@ def variance(times):
return round(sum(((i - mean) ** 2.0 for i in times)) / (len(times) - 1), 3)
async def wait_for_done(conf, uri, timeout):
name = uri.split("#")[0]
last_complete = 0
hang_count = 0
async def wait_for_done(conf, claim_name, timeout):
blobs_completed, last_completed = 0, time.perf_counter()
while True:
files = await daemon_rpc(conf, "file_list", claim_name=name)
file = files[0]
file = (await daemon_rpc(conf, "file_list", claim_name=claim_name))[0]
if file['status'] in ['finished', 'stopped']:
return True, file['blobs_completed'], file['blobs_in_stream']
if last_complete < int(file['blobs_completed']):
hang_count = 0
last_complete = int(file['blobs_completed'])
else:
hang_count += 1
await asyncio.sleep(1.0)
if hang_count > timeout:
elif blobs_completed < int(file['blobs_completed']):
blobs_completed, last_completed = int(file['blobs_completed']), time.perf_counter()
elif (time.perf_counter() - last_completed) > timeout:
return False, file['blobs_completed'], file['blobs_in_stream']
await asyncio.sleep(1.0)
async def main(uris=None, cmd_args=None):
if not uris:
uris = await get_frontpage_uris()
async def main(cmd_args=None):
print('Time to first byte started using parameters:')
for key, value in vars(cmd_args).items():
print(f"{key}: {value}")
conf = Config()
url_to_claim = {}
try:
await daemon_rpc(conf, 'status')
for page in range(1, cmd_args.download_pages + 1):
start = time.perf_counter()
kwargs = {
'page': page,
# 'claim_type': 'stream',
'order_by': ['trending_global', 'trending_mixed'],
'no_totals': True
}
# if not cmd_args.allow_fees:
# kwargs['fee_amount'] = 0
response = await daemon_rpc(
conf, 'claim_search', **kwargs
)
if 'error' in response or not response.get('items'):
print(f'Error getting claim list page {page}:')
print(response)
return 1
else:
url_to_claim.update({
claim['permanent_url']: claim for claim in response['items']
})
print(f'Claim search page {page} took: {time.perf_counter() - start}')
except (ClientConnectorError, ConnectionError):
print("Could not connect to daemon")
return 1
print(f"Checking {len(uris)} uris from the front page")
print("**********************************************")
resolvable = []
async def __resolve(name):
resolved = await daemon_rpc(conf, 'resolve', urls=[name])
if 'error' not in resolved.get(name, {}):
if ("fee" not in resolved[name]['claim']['value']) or cmd_args.allow_fees:
resolvable.append(name)
else:
print(f"{name} has a fee, skipping it")
else:
print(f"failed to resolve {name}: {resolved[name]['error']}")
await asyncio.gather(*(__resolve(name) for name in uris))
print(f"attempting to download {len(resolvable)}/{len(uris)} frontpage streams")
print(f"Attempting to download {len(url_to_claim)} claim_search streams")
first_byte_times = []
download_speeds = []
download_successes = []
failed_to_start = []
download_failures = []
failed_to = {}
for uri in resolvable:
await daemon_rpc(conf, 'file_delete', delete_from_download_dir=True, claim_name=parse_lbry_uri(uri).name)
await asyncio.gather(*(
daemon_rpc(conf, 'file_delete', delete_from_download_dir=True, claim_name=claim['name'])
for claim in url_to_claim.values() if not cmd_args.keep_files
))
for i, uri in enumerate(resolvable):
start = time.time()
try:
await daemon_rpc(conf, 'get', uri=uri, save_file=True)
first_byte = time.time()
first_byte_times.append(first_byte - start)
print(f"{i + 1}/{len(resolvable)} - {first_byte - start} {uri}")
if not cmd_args.head_blob_only:
downloaded, amount_downloaded, blobs_in_stream = await wait_for_done(
conf, uri, cmd_args.stall_download_timeout
)
if downloaded:
download_successes.append(uri)
else:
download_failures.append(uri)
mbs = round((blobs_in_stream * (MAX_BLOB_SIZE - 1)) / (time.time() - start) / 1000000, 2)
download_speeds.append(mbs)
print(f"downloaded {amount_downloaded}/{blobs_in_stream} blobs for {uri} at "
f"{mbs}mb/s")
except Exception as e:
print(f"{i + 1}/{len(uris)} - failed to start {uri}: {e}")
failed_to_start.append(uri)
for i, (url, claim) in enumerate(url_to_claim.items()):
start = time.perf_counter()
response = await daemon_rpc(conf, 'get', uri=url, save_file=not cmd_args.head_blob_only)
if 'error' in response:
print(f"{i + 1}/{len(url_to_claim)} - failed to start {url}: {response['error']}")
failed_to[url] = 'start'
if cmd_args.exit_on_error:
return
if cmd_args.delete_after_download or cmd_args.head_blob_only:
await daemon_rpc(conf, 'file_delete', delete_from_download_dir=True, claim_name=parse_lbry_uri(uri).name)
continue
first_byte = time.perf_counter()
first_byte_times.append(first_byte - start)
print(f"{i + 1}/{len(url_to_claim)} - {first_byte - start} {url}")
if not cmd_args.head_blob_only:
downloaded, amount_downloaded, blobs_in_stream = await wait_for_done(
conf, claim['name'], cmd_args.stall_download_timeout
)
if downloaded:
download_successes.append(url)
else:
failed_to[url] = 'finish'
mbs = round((blobs_in_stream * (MAX_BLOB_SIZE - 1)) / (time.perf_counter() - start) / 1000000, 2)
download_speeds.append(mbs)
print(f"downloaded {amount_downloaded}/{blobs_in_stream} blobs for {url} at "
f"{mbs}mb/s")
if not cmd_args.keep_files:
await daemon_rpc(conf, 'file_delete', delete_from_download_dir=True, claim_name=claim['name'])
await asyncio.sleep(0.1)
print("**********************************************")
result = f"Started {len(first_byte_times)} of {len(resolvable)} attempted front page streams\n" \
f"Worst first byte time: {round(max(first_byte_times), 2)}\n" \
f"Best first byte time: {round(min(first_byte_times), 2)}\n" \
f"95% confidence time-to-first-byte: {confidence(first_byte_times, 1.984)}s\n" \
f"99% confidence time-to-first-byte: {confidence(first_byte_times, 2.626)}s\n" \
f"Variance: {variance(first_byte_times)}\n"
if not cmd_args.head_blob_only:
result += f"Downloaded {len(download_successes)}/{len(resolvable)}\n" \
result = f"Started {len(first_byte_times)} of {len(url_to_claim)} attempted front page streams\n"
if first_byte_times:
result += f"Worst first byte time: {round(max(first_byte_times), 2)}\n" \
f"Best first byte time: {round(min(first_byte_times), 2)}\n" \
f"95% confidence time-to-first-byte: {confidence(first_byte_times, 1.984)}s\n" \
f"99% confidence time-to-first-byte: {confidence(first_byte_times, 2.626)}s\n" \
f"Variance: {variance(first_byte_times)}\n"
if download_successes:
result += f"Downloaded {len(download_successes)}/{len(url_to_claim)}\n" \
f"Best stream download speed: {round(max(download_speeds), 2)}mb/s\n" \
f"Worst stream download speed: {round(min(download_speeds), 2)}mb/s\n" \
f"95% confidence download speed: {confidence(download_speeds, 1.984, False)}mb/s\n" \
f"99% confidence download speed: {confidence(download_speeds, 2.626, False)}mb/s\n"
if failed_to_start:
result += "\nFailed to start:" + "\n".join([f for f in failed_to_start])
if download_failures:
result += "\nFailed to finish:" + "\n".join([f for f in download_failures])
for reason in ('start', 'finish'):
failures = [url for url, why in failed_to.items() if reason == why]
if failures:
result += f"\nFailed to {reason}:\n" + "\n".join(failures)
print(result)
webhook = os.environ.get('TTFB_SLACK_TOKEN', None)
if webhook:
await report_to_slack(result, webhook)
return 0
if __name__ == "__main__":
parser = argparse.ArgumentParser()
#parser.add_argument("--data_dir")
#parser.add_argument("--wallet_dir")
#parser.add_argument("--download_directory")
parser.add_argument("--allow_fees", action='store_true')
parser.add_argument("--exit_on_error", action='store_true')
parser.add_argument("--stall_download_timeout", default=10, type=int)
parser.add_argument("--delete_after_download", action='store_true')
parser.add_argument("--stall_download_timeout", default=0, type=int)
parser.add_argument("--keep_files", action='store_true')
parser.add_argument("--head_blob_only", action='store_true')
asyncio.run(main(cmd_args=parser.parse_args()))
parser.add_argument("--download_pages", type=int, default=10)
sys.exit(asyncio.run(main(cmd_args=parser.parse_args())) or 0)
+87 -12
View File
@@ -1,3 +1,4 @@
import time
from math import ceil
from aiohttp import web
@@ -14,6 +15,19 @@ class MockedCommentServer:
'INVALID_METHOD': {'code': -32604, 'message': 'The Requested method does not exist'}
}
COMMENT_SCHEMA = {
'comment': None,
'comment_id': None,
'claim_id': None,
'parent_id': None,
'channel_name': None,
'channel_id': None,
'signature': None,
'signing_ts': None,
'timestamp': None,
'channel_url': None,
}
def __init__(self, port=2903):
self.port = port
self.app = web.Application(debug=True)
@@ -23,13 +37,43 @@ class MockedCommentServer:
self.comments = []
self.comment_id = 0
def create_comment(self, **comment):
@classmethod
def _create_comment(cls, **kwargs):
schema = cls.COMMENT_SCHEMA.copy()
schema.update(**kwargs)
return schema
@staticmethod
def clean(d: dict):
return {k: v for k, v in d.items() if v}
def create_comment(self, channel_name=None, channel_id=None, **kwargs):
self.comment_id += 1
comment['comment_id'] = self.comment_id
if 'channel_id' in comment:
comment['channel_url'] = 'lbry://' + comment['channel_name'] + '#' + comment['channel_id']
comment_id = self.comment_id
channel_url = 'lbry://' + channel_name + '#' + channel_id if channel_id else None
comment = self._create_comment(
comment_id=str(comment_id),
channel_name=channel_name,
channel_id=channel_id,
channel_url=channel_url,
timestamp=str(int(time.time())),
**kwargs
)
self.comments.append(comment)
return comment
return self.clean(comment)
def delete_comment(self, comment_id: int, channel_id: str, **kwargs):
deleted = False
try:
if 0 <= comment_id <= len(self.comments) and self.comments[comment_id - 1]['channel_id'] == channel_id:
self.comments.pop(comment_id - 1)
deleted = True
finally:
return {
str(comment_id): {
'deleted': deleted
}
}
def get_claim_comments(self, page=1, page_size=50, **kwargs):
return {
@@ -37,22 +81,36 @@ class MockedCommentServer:
'page_size': page_size,
'total_pages': ceil(len(self.comments)/page_size),
'total_items': len(self.comments),
'items': (self.comments[::-1])[(page - 1) * page_size: page * page_size]
'items': [self.clean(c) for c in (self.comments[::-1])[(page - 1) * page_size: page * page_size]]
}
def get_comment_channel_by_id(self, comment_id: int, **kwargs):
comment = self.comments[comment_id - 1]
return {
'channel_id': comment.get('channel_id'),
'channel_name': comment.get('channel_name')
}
methods = {
'get_claim_comments': get_claim_comments,
'create_comment': create_comment,
'delete_comment': delete_comment,
'get_channel_from_comment_id': get_comment_channel_by_id,
}
def process_json(self, body) -> dict:
response = {'jsonrpc': '2.0', 'id': body['id']}
if body['method'] in self.methods:
params = body.get('params', {})
result = self.methods[body['method']](self, **params)
response['result'] = result
else:
response['error'] = self.ERRORS['INVALID_METHOD']
try:
if body['method'] in self.methods:
params = body.get('params', {})
if 'comment_id' in params and type(params['comment_id']) is str:
params['comment_id'] = int(params['comment_id'])
result = self.methods[body['method']](self, **params)
response['result'] = result
else:
response['error'] = self.ERRORS['INVALID_METHOD']
except Exception:
response['error'] = self.ERRORS['UNKNOWN']
return response
async def start(self):
@@ -173,3 +231,20 @@ class CommentCommands(CommandTestCase):
is_channel_signature_valid=True
)
self.assertIs(len(signed_comment_list['items']), 28)
async def test04_comment_abandons(self):
rswanson = (await self.channel_create('@RonSwanson'))['outputs'][0]
stream = (await self.stream_create('Pawnee Town Hall of Fame by Leslie Knope'))['outputs'][0]
comment = await self.daemon.jsonrpc_comment_create(
comment='KNOPE! WHAT DID I TELL YOU ABOUT PUTTING MY INFORMATION UP LIKE THAT',
claim_id=stream['claim_id'],
channel_id=rswanson['claim_id']
)
self.assertIn('signature', comment)
deleted = await self.daemon.jsonrpc_comment_abandon(comment['comment_id'])
self.assertIn(comment['comment_id'], deleted)
self.assertTrue(deleted[comment['comment_id']]['deleted'])
deleted = await self.daemon.jsonrpc_comment_abandon(comment['comment_id'])
self.assertIn(comment['comment_id'], deleted)
self.assertFalse(deleted[comment['comment_id']]['deleted'])
@@ -1,9 +1,11 @@
import asyncio
import os
from torba.client.basenetwork import ClientSession
from torba.orchstr8 import Conductor
from torba.testcase import IntegrationTestCase
import lbry.wallet
from lbry.testcase import CommandTestCase
from lbry.extras.daemon.Components import HeadersComponent
from torba.client.basenetwork import ClientSession
from torba.testcase import IntegrationTestCase
class TestSessionBloat(IntegrationTestCase):
@@ -29,3 +31,22 @@ class TestSessionBloat(IntegrationTestCase):
await session.send_request('server.banner', ())
self.assertTrue(session.is_closing())
self.assertEqual(len(self.conductor.spv_node.server.session_mgr.sessions), 0)
class TestHeadersComponent(CommandTestCase):
LEDGER = lbry.wallet
async def asyncSetUp(self):
await super().asyncSetUp()
self.component_manager = self.daemon.component_manager
self.component_manager.conf.blockchain_name = 'lbrycrd_main'
self.headers_component = HeadersComponent(self.component_manager)
async def test_cant_reach_host(self):
HeadersComponent.HEADERS_URL = 'notthere/'
os.unlink(self.headers_component.headers.path)
# test is that this doesnt raise
await self.headers_component.start()
self.assertTrue(self.component_manager.get_components_status()['blockchain_headers'])
self.assertEqual(await self.headers_component.get_status(), {})
@@ -1,5 +1,5 @@
from torba.testcase import AsyncioTestCase
import hashlib
from lbry.extras.daemon.comment_client import sign_comment
from lbry.extras.daemon.comment_client import is_comment_signed_by_channel
@@ -14,7 +14,8 @@ class TestSigningComments(AsyncioTestCase):
'claim_id': claim.claim_id,
'channel_name': channel.claim_name,
'channel_id': channel.claim_id,
'comment': comment
'comment': comment,
'comment_id': hashlib.sha256(comment.encode()).hexdigest()
}
def test01_successful_create_sign_and_validate_comment(self):
@@ -31,3 +32,28 @@ class TestSigningComments(AsyncioTestCase):
comment = self.create_claim_comment_body('Woahh This is Sick!! Shout out 2 my boy Tommy H', stream, pdiddy)
sign_comment(comment, channel2)
self.assertFalse(is_comment_signed_by_channel(comment, pdiddy))
def test03_successful_sign_abandon_comment(self):
rswanson = get_channel('@RonSwanson')
dsilver = get_stream('Welcome to the Pawnee, and give a big round for Ron Swanson, AKA Duke Silver')
comment_body = self.create_claim_comment_body('COMPUTER, DELETE ALL VIDEOS OF RON.', dsilver, rswanson)
sign_comment(comment_body, rswanson, abandon=True)
self.assertTrue(is_comment_signed_by_channel(comment_body, rswanson, abandon=True))
def test04_invalid_signature(self):
rswanson = get_channel('@RonSwanson')
jeanralphio = get_channel('@JeanRalphio')
chair = get_stream('This is a nice chair. I made it with Mahogany wood and this electric saw')
chair_comment = self.create_claim_comment_body(
'Hah. You use an electric saw? Us swansons have been making chairs with handsaws just three after birth.',
chair,
rswanson
)
sign_comment(chair_comment, rswanson)
self.assertTrue(is_comment_signed_by_channel(chair_comment, rswanson))
self.assertFalse(is_comment_signed_by_channel(chair_comment, jeanralphio))
fake_abandon_signal = chair_comment.copy()
sign_comment(fake_abandon_signal, jeanralphio, abandon=True)
self.assertFalse(is_comment_signed_by_channel(fake_abandon_signal, rswanson, abandon=True))
self.assertFalse(is_comment_signed_by_channel(fake_abandon_signal, jeanralphio, abandon=True))
@@ -109,7 +109,6 @@ class BasicHeadersTests(BitcoinHeadersTestCase):
# corrupt by appending
headers.io.seek(block_bytes(len(headers)))
headers.io.write(b"appending")
headers._size = None
await headers.repair()
self.assertEqual(headers.height, 1499)
await headers.connect(len(headers), self.get_bytes(block_bytes(3001 - 1500), after=block_bytes(1500)))
+7 -5
View File
@@ -115,7 +115,7 @@ class BaseHeaders:
# .seek()/.write()/.truncate() might also .flush() when needed
# the goal here is mainly to ensure we're definitely flush()'ing
self.io.flush()
self._size = None
self._size = self.io.tell() // self.header_size
added += written
if bail:
break
@@ -170,10 +170,12 @@ class BaseHeaders:
async def repair(self):
previous_header_hash = fail = None
self.io.seek(0)
batch_size = 10000
batch_size = 36
for start_height in range(0, self.height, batch_size):
self.io.seek(self.header_size * start_height)
headers = self.io.read(self.header_size*batch_size)
if len(headers) % self.header_size != 0:
headers = headers[:(len(headers) // self.header_size) * self.header_size]
for header_hash, header in self._iterate_headers(start_height, headers):
height = header['block_height']
if height:
@@ -184,7 +186,7 @@ class BaseHeaders:
fail = True
if fail:
log.warning("Header file corrupted at height %s, truncating it.", height - 1)
self.io.seek((height - 1) * self.header_size, os.SEEK_SET)
self.io.seek(max(0, (height - 1)) * self.header_size, os.SEEK_SET)
self.io.truncate()
self.io.flush()
self._size = None
@@ -205,7 +207,7 @@ class BaseHeaders:
end = min(len(headers), end + self.chunk_size * self.header_size)
def _iterate_headers(self, height: int, headers: bytes) -> Iterator[Tuple[bytes, dict]]:
assert len(headers) % self.header_size == 0
assert len(headers) % self.header_size == 0, len(headers)
for idx in range(len(headers) // self.header_size):
start, end = idx * self.header_size, (idx + 1) * self.header_size
header = headers[start:end]