Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Specify API-level search timeout to reduce hard timeout errors. #81

Merged
merged 4 commits into from
Sep 12, 2022
Merged

Specify API-level search timeout to reduce hard timeout errors. #81

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
a3b7795
Specify API-level search timeout to reduce hard timeout errors.
moodyjon Aug 17, 2022
28c711e
Report API-level search timeouts in interrupt_count_metric.
moodyjon Aug 18, 2022
4c5ca2d
Revert "Report API-level search timeouts in interrupt_count_metric."
moodyjon Aug 24, 2022
08931a1
Updates for review comments. Implement timeout counter
moodyjon Aug 24, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 11 additions & 5 deletions hub/herald/search.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,6 +10,7 @@
from hub.common import LRUCache, IndexVersionMismatch, INDEX_DEFAULT_SETTINGS, expand_query, expand_result
from hub.db.common import ResolveResult
if TYPE_CHECKING:
from prometheus_client import Counter as PrometheusCounter
from hub.db import SecondaryDB


Expand All @@ -29,9 +30,10 @@ class SearchIndex:
VERSION = 1

def __init__(self, hub_db: 'SecondaryDB', index_prefix: str, search_timeout=3.0, elastic_host='localhost',
elastic_port=9200):
elastic_port=9200, timeout_counter: Optional['PrometheusCounter'] = None):
self.hub_db = hub_db
self.search_timeout = search_timeout
self.timeout_counter: Optional['PrometheusCounter'] = timeout_counter
self.sync_timeout = 600 # wont hit that 99% of the time, but can hit on a fresh import
self.search_client: Optional[AsyncElasticsearch] = None
self.sync_client: Optional[AsyncElasticsearch] = None
Expand Down Expand Up @@ -59,7 +61,7 @@ async def start(self) -> bool:
return False
hosts = [{'host': self._elastic_host, 'port': self._elastic_port}]
self.sync_client = AsyncElasticsearch(hosts, timeout=self.sync_timeout)
self.search_client = AsyncElasticsearch(hosts, timeout=self.search_timeout)
self.search_client = AsyncElasticsearch(hosts, timeout=self.search_timeout+1)
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

What was the +1 for? Can this be removed?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

If the transport timeout is less than (or possibly equal to) the API timeout, then we will likely continue to get hard timeout errors. Bumping it by 1 is a way of pushing it out beyond the API timeout. The API timeout should trigger first, and we get partial results.

I could make the API timeout some fraction of the transport timeout...

while True:
try:
await self.sync_client.cluster.health(wait_for_status='yellow')
Expand Down Expand Up @@ -205,10 +207,14 @@ async def search_ahead(self, **kwargs):
reordered_hits = cache_item.result
else:
query = expand_query(**kwargs)
search_hits = deque((await self.search_client.search(
es_resp = await self.search_client.search(
query, index=self.index, track_total_hits=False,
timeout=f'{int(1000*self.search_timeout)}ms',
_source_includes=['_id', 'channel_id', 'reposted_claim_id', 'creation_height']
))['hits']['hits'])
)
search_hits = deque(es_resp['hits']['hits'])
if self.timeout_counter and es_resp['timed_out']:
self.timeout_counter.inc()
if remove_duplicates:
search_hits = self.__remove_duplicates(search_hits)
if per_channel_per_page > 0:
Expand All @@ -235,7 +241,7 @@ def __remove_duplicates(self, search_hits: deque) -> deque:
dropped.add(hit['_id'])
return deque(hit for hit in search_hits if hit['_id'] not in dropped)

def __search_ahead(self, search_hits: list, page_size: int, per_channel_per_page: int):
def __search_ahead(self, search_hits: deque, page_size: int, per_channel_per_page: int) -> list:
reordered_hits = []
channel_counters = Counter()
next_page_hits_maybe_check_later = deque()
Expand Down
3 changes: 2 additions & 1 deletion hub/herald/session.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -210,7 +210,8 @@ def __init__(self, env: 'ServerEnv', db: 'SecondaryDB', mempool: 'HubMemPool',
# Search index
self.search_index = SearchIndex(
self.db, self.env.es_index_prefix, self.env.database_query_timeout,
elastic_host=env.elastic_host, elastic_port=env.elastic_port
elastic_host=env.elastic_host, elastic_port=env.elastic_port,
timeout_counter=self.interrupt_count_metric
)
self.running = False
# hashX: List[int]
Expand Down