vbatts/quay
Archived
1
0
Fork 0
Code Issues Pull requests Releases Wiki Activity

Temporarily change to storing logs in a new LogEntry2 table

Browse source 
This will prevent us from running out of auto-incrementing ID values until such time as we can upgrade to peewee 3 and change the field type to a BigInt

Fixes https://jira.coreos.com/browse/QUAY-943
This commit is contained in:
Joseph Schorr 2018-05-18 12:54:38 -04:00
parent 66b4e45929
commit a007332d4c
13 changed files with 201 additions and 113 deletions
Download patch file Download diff file Expand all files Collapse all files

75
data/model/log.py
View file

@ -7,7 +7,7 @@ from datetime import datetime, timedelta
from cachetools import lru_cache
import data
from data.database import LogEntry, LogEntryKind, User, RepositoryActionCount, db
from data.database import LogEntry, LogEntry2, LogEntryKind, User, RepositoryActionCount, db
from data.model import config, user, DataModelException
logger = logging.getLogger(__name__)
@ -16,27 +16,29 @@ ACTIONS_ALLOWED_WITHOUT_AUDIT_LOGGING = ['pull_repo']
def _logs_query(selections, start_time, end_time, performer=None, repository=None, namespace=None,
ignore=None):
joined = (LogEntry.select(*selections).switch(LogEntry)
.where(LogEntry.datetime >= start_time, LogEntry.datetime < end_time))
ignore=None, model=LogEntry):
""" Returns a query for selecting logs from the table, with various options and filters. """
# TODO(LogMigrate): Remove the branch once we're back on LogEntry only.
joined = (model.select(*selections).switch(model)
.where(model.datetime >= start_time, model.datetime < end_time))
if repository:
joined = joined.where(LogEntry.repository == repository)
joined = joined.where(model.repository == repository)
if performer:
joined = joined.where(LogEntry.performer == performer)
joined = joined.where(model.performer == performer)
if namespace and not repository:
namespace_user = user.get_user_or_org(namespace)
if namespace_user is None:
raise DataModelException('Invalid namespace requested')
joined = joined.where(LogEntry.account == namespace_user.id)
joined = joined.where(model.account == namespace_user.id)
if ignore:
kind_map = get_log_entry_kinds()
ignore_ids = [kind_map[kind_name] for kind_name in ignore]
joined = joined.where(~(LogEntry.kind << ignore_ids))
joined = joined.where(~(model.kind << ignore_ids))
return joined
@ -57,29 +59,35 @@ def _get_log_entry_kind(name):
def get_aggregated_logs(start_time, end_time, performer=None, repository=None, namespace=None,
ignore=None):
date = db.extract_date('day', LogEntry.datetime)
selections = [LogEntry.kind, date.alias('day'), fn.Count(LogEntry.id).alias('count')]
query = _logs_query(selections, start_time, end_time, performer, repository, namespace, ignore)
return query.group_by(date, LogEntry.kind)
ignore=None, model=LogEntry):
""" Returns the count of logs, by kind and day, for the logs matching the given filters. """
# TODO(LogMigrate): Remove the branch once we're back on LogEntry only.
date = db.extract_date('day', model.datetime)
selections = [model.kind, date.alias('day'), fn.Count(model.id).alias('count')]
query = _logs_query(selections, start_time, end_time, performer, repository, namespace, ignore,
model=model)
return query.group_by(date, model.kind)
def get_logs_query(start_time, end_time, performer=None, repository=None, namespace=None,
ignore=None):
ignore=None, model=LogEntry):
""" Returns the logs matching the given filters. """
# TODO(LogMigrate): Remove the branch once we're back on LogEntry only.
Performer = User.alias()
Account = User.alias()
selections = [LogEntry, Performer]
selections = [model, Performer]
if namespace is None and repository is None:
selections.append(Account)
query = _logs_query(selections, start_time, end_time, performer, repository, namespace, ignore)
query = (query.switch(LogEntry).join(Performer, JOIN_LEFT_OUTER,
on=(LogEntry.performer == Performer.id).alias('performer')))
query = _logs_query(selections, start_time, end_time, performer, repository, namespace, ignore,
model=model)
query = (query.switch(model).join(Performer, JOIN_LEFT_OUTER,
on=(model.performer == Performer.id).alias('performer')))
if namespace is None and repository is None:
query = (query.switch(LogEntry).join(Account, JOIN_LEFT_OUTER,
on=(LogEntry.account == Account.id).alias('account')))
query = (query.switch(model).join(Account, JOIN_LEFT_OUTER,
on=(model.account == Account.id).alias('account')))
return query
@ -93,6 +101,7 @@ def _json_serialize(obj):
def log_action(kind_name, user_or_organization_name, performer=None, repository=None, ip=None,
metadata={}, timestamp=None):
""" Logs an entry in the LogEntry2 table. """
if not timestamp:
timestamp = datetime.today()
@ -123,7 +132,7 @@ def log_action(kind_name, user_or_organization_name, performer=None, repository=
}
try:
LogEntry.create(**log_data)
LogEntry2.create(**log_data)
except PeeweeException as ex:
strict_logging_disabled = config.app_config.get('ALLOW_PULLS_WITHOUT_STRICT_LOGGING')
if strict_logging_disabled and kind_name in ACTIONS_ALLOWED_WITHOUT_AUDIT_LOGGING:
@ -132,39 +141,47 @@ def log_action(kind_name, user_or_organization_name, performer=None, repository=
raise
def get_stale_logs_start_id():
def get_stale_logs_start_id(model):
""" Gets the oldest log entry. """
# TODO(LogMigrate): Remove the branch once we're back on LogEntry only.
try:
return (LogEntry.select(LogEntry.id).order_by(LogEntry.id).limit(1).tuples())[0][0]
return (model.select(model.id).order_by(model.id).limit(1).tuples())[0][0]
except IndexError:
return None
def get_stale_logs_cutoff_id(cutoff_date):
def get_stale_logs_cutoff_id(cutoff_date, model):
""" Gets the most recent ID created before the cutoff_date. """
# TODO(LogMigrate): Remove the branch once we're back on LogEntry only.
try:
return (LogEntry.select(fn.Max(LogEntry.id)).where(LogEntry.datetime <= cutoff_date)
return (model.select(fn.Max(model.id)).where(model.datetime <= cutoff_date)
.tuples())[0][0]
except IndexError:
return None
def get_stale_logs(start_id, end_id):
def get_stale_logs(start_id, end_id, model):
""" Returns all the logs with IDs between start_id and end_id inclusively. """
return LogEntry.select().where((LogEntry.id >= start_id), (LogEntry.id <= end_id))
# TODO(LogMigrate): Remove the branch once we're back on LogEntry only.
return model.select().where((model.id >= start_id), (model.id <= end_id))
def delete_stale_logs(start_id, end_id):
def delete_stale_logs(start_id, end_id, model):
""" Deletes all the logs with IDs between start_id and end_id. """
LogEntry.delete().where((LogEntry.id >= start_id), (LogEntry.id <= end_id)).execute()
# TODO(LogMigrate): Remove the branch once we're back on LogEntry only.
model.delete().where((model.id >= start_id), (model.id <= end_id)).execute()
def get_repository_action_counts(repo, start_date):
""" Returns the daily aggregated action counts for the given repository, starting at the given
start date.
"""
return RepositoryActionCount.select().where(RepositoryActionCount.repository == repo,
RepositoryActionCount.date >= start_date)
def get_repositories_action_sums(repository_ids):
""" Returns a map from repository ID to total actions within that repository in the last week. """
if not repository_ids:
return {}