def forwards_dist(self) -> Sequence[operations.SqlOperation]:
return [
operations.AddColumn(
storage_set=StorageSetKey.EVENTS,
table_name="errors_dist",
column=Column("hierarchical_hashes", Array(UUID())),
after="primary_hash",
),
operations.AddColumn(
storage_set=StorageSetKey.EVENTS,
table_name="sentry_dist",
column=Column("hierarchical_hashes", Array(FixedString(32)),),
after="primary_hash",
),
]
def __init__(self) -> None:
self.__common_columns = ColumnSet(
[
("event_id", FixedString(32)),
("project_id", UInt(64)),
("type", Nullable(String())),
("timestamp", DateTime()),
("platform", Nullable(String())),
("environment", Nullable(String())),
("release", Nullable(String())),
("dist", Nullable(String())),
("user", Nullable(String())),
("transaction", Nullable(String())),
("message", Nullable(String())),
("title", Nullable(String())),
# User
("user_id", Nullable(String())),
("username", Nullable(String())),
("email", Nullable(String())),
("ip_address", Nullable(String())),
# SDK
("sdk_name", Nullable(String())),
("sdk_version", Nullable(String())),
# geo location context
("geo_country_code", Nullable(String())),
("geo_region", Nullable(String())),
("geo_city", Nullable(String())),
("http_method", Nullable(String())),
("http_referer", Nullable(String())),
# Other tags and context
("tags", Nested([("key", String()), ("value", String())])),
("contexts", Nested([("key", String()), ("value", String())])),
]
)
self.__events_columns = ColumnSet(
[
("group_id", Nullable(UInt(64))),
("primary_hash", Nullable(FixedString(32))),
# Promoted tags
("level", Nullable(String())),
("logger", Nullable(String())),
("server_name", Nullable(String())),
("site", Nullable(String())),
("url", Nullable(String())),
("search_message", Nullable(String())),
("location", Nullable(String())),
("culprit", Nullable(String())),
("received", Nullable(DateTime())),
("sdk_integrations", Nullable(Array(String()))),
("version", Nullable(String())),
# exception interface
(
"exception_stacks",
Nested(
[
("type", Nullable(String())),
("value", Nullable(String())),
("mechanism_type", Nullable(String())),
("mechanism_handled", Nullable(UInt(8))),
]
),
),
(
"exception_frames",
Nested(
[
("abs_path", Nullable(String())),
("filename", Nullable(String())),
("package", Nullable(String())),
("module", Nullable(String())),
("function", Nullable(String())),
("in_app", Nullable(UInt(8))),
("colno", Nullable(UInt(32))),
("lineno", Nullable(UInt(32))),
("stack_level", UInt(16)),
]
),
),
("modules", Nested([("name", String()), ("version", String())])),
]
)
self.__transactions_columns = ColumnSet(
[
("trace_id", Nullable(UUID())),
("span_id", Nullable(UInt(64))),
("transaction_hash", Nullable(UInt(64))),
("transaction_op", Nullable(String())),
("transaction_status", Nullable(UInt(8))),
("duration", Nullable(UInt(32))),
(
"measurements",
Nested([("key", LowCardinality(String())), ("value", Float(64))]),
),
]
)
events_storage = get_storage(StorageKey.EVENTS)
events_ro_storage = get_storage(StorageKey.EVENTS_RO)
transactions_storage = get_storage(StorageKey.TRANSACTIONS)
self.__time_group_columns: Mapping[str, str] = {}
self.__time_parse_columns = ("timestamp",)
super().__init__(
storages=[events_storage, transactions_storage],
query_plan_builder=SelectedStorageQueryPlanBuilder(
selector=DiscoverQueryStorageSelector(
events_table=events_storage,
events_ro_table=events_ro_storage,
abstract_events_columns=self.__events_columns,
transactions_table=transactions_storage,
abstract_transactions_columns=self.__transactions_columns,
),
),
abstract_column_set=(
self.__common_columns
+ self.__events_columns
+ self.__transactions_columns
),
writable_storage=None,
)
("device_brand", Nullable(String())),
("device_locale", Nullable(String())),
("device_uuid", Nullable(String())),
("device_model_id", Nullable(String())),
("device_arch", Nullable(String())),
("device_battery_level", Nullable(Float(32))),
("device_orientation", Nullable(String())),
("device_simulator", Nullable(UInt(8))),
("device_online", Nullable(UInt(8))),
("device_charging", Nullable(UInt(8))),
]
)
required_columns = ColumnSet(
[
("event_id", FixedString(32)),
("project_id", UInt(64)),
("group_id", UInt(64)),
("timestamp", DateTime()),
("deleted", UInt(8)),
("retention_days", UInt(16)),
]
)
all_columns = (
required_columns
+ [
# required for non-deleted
("platform", Nullable(String())),
("message", Nullable(String())),
("primary_hash", Nullable(FixedString(32))),
("device_brand", String(Modifiers(nullable=True))),
("device_locale", String(Modifiers(nullable=True))),
("device_uuid", String(Modifiers(nullable=True))),
("device_model_id", String(Modifiers(nullable=True))),
("device_arch", String(Modifiers(nullable=True))),
("device_battery_level", Float(32, Modifiers(nullable=True))),
("device_orientation", String(Modifiers(nullable=True))),
("device_simulator", UInt(8, Modifiers(nullable=True))),
("device_online", UInt(8, Modifiers(nullable=True))),
("device_charging", UInt(8, Modifiers(nullable=True))),
]
)
required_columns = ColumnSet(
[
("event_id", FixedString(32)),
("project_id", UInt(64)),
("group_id", UInt(64)),
("timestamp", DateTime()),
("deleted", UInt(8)),
("retention_days", UInt(16)),
]
)
all_columns = (
required_columns
+ [
# required for non-deleted
("platform", String(Modifiers(nullable=True))),
("message", String(Modifiers(nullable=True))),
("primary_hash", FixedString(32, Modifiers(nullable=True))),
("os_kernel_version", String(Modifiers(nullable=True))),
("device_name", String(Modifiers(nullable=True))),
("device_brand", String(Modifiers(nullable=True))),
("device_locale", String(Modifiers(nullable=True))),
("device_uuid", String(Modifiers(nullable=True))),
("device_model_id", String(Modifiers(nullable=True))),
("device_arch", String(Modifiers(nullable=True))),
("device_battery_level", Float(32, Modifiers(nullable=True))),
("device_orientation", String(Modifiers(nullable=True))),
("device_simulator", UInt(8, Modifiers(nullable=True))),
("device_online", UInt(8, Modifiers(nullable=True))),
("device_charging", UInt(8, Modifiers(nullable=True))),
])
required_columns = ColumnSet([
("event_id", FixedString(32)),
("project_id", UInt(64)),
("group_id", UInt(64)),
("timestamp", DateTime()),
("deleted", UInt(8)),
("retention_days", UInt(16)),
])
all_columns = (
required_columns + [
# required for non-deleted
("platform", String(Modifiers(nullable=True))),
("message", String(Modifiers(nullable=True))),
("primary_hash", FixedString(32, Modifiers(nullable=True))),
("hierarchical_hashes", Array(FixedString(32))),
("received", DateTime(Modifiers(nullable=True))),
def __init__(self) -> None:
metadata_columns = ColumnSet([
# optional stream related data
("offset", Nullable(UInt(64))),
("partition", Nullable(UInt(16))),
])
promoted_tag_columns = ColumnSet([
# These are the classic tags, they are saved in Snuba exactly as they
# appear in the event body.
("level", Nullable(String())),
("logger", Nullable(String())),
("server_name", Nullable(String())), # future name: device_id?
("transaction", Nullable(String())),
("environment", Nullable(String())),
("sentry:release", Nullable(String())),
("sentry:dist", Nullable(String())),
("sentry:user", Nullable(String())),
("site", Nullable(String())),
("url", Nullable(String())),
])
promoted_context_tag_columns = ColumnSet([
# These are promoted tags that come in in `tags`, but are more closely
# related to contexts. To avoid naming confusion with Clickhouse nested
# columns, they are stored in the database with s/./_/
# promoted tags
("app_device", Nullable(String())),
("device", Nullable(String())),
("device_family", Nullable(String())),
("runtime", Nullable(String())),
("runtime_name", Nullable(String())),
("browser", Nullable(String())),
("browser_name", Nullable(String())),
("os", Nullable(String())),
("os_name", Nullable(String())),
("os_rooted", Nullable(UInt(8))),
])
promoted_context_columns = ColumnSet([
("os_build", Nullable(String())),
("os_kernel_version", Nullable(String())),
("device_name", Nullable(String())),
("device_brand", Nullable(String())),
("device_locale", Nullable(String())),
("device_uuid", Nullable(String())),
("device_model_id", Nullable(String())),
("device_arch", Nullable(String())),
("device_battery_level", Nullable(Float(32))),
("device_orientation", Nullable(String())),
("device_simulator", Nullable(UInt(8))),
("device_online", Nullable(UInt(8))),
("device_charging", Nullable(UInt(8))),
])
required_columns = ColumnSet([
("event_id", FixedString(32)),
("project_id", UInt(64)),
("group_id", UInt(64)),
("timestamp", DateTime()),
("deleted", UInt(8)),
("retention_days", UInt(16)),
])
all_columns = (
required_columns + [
# required for non-deleted
("platform", Nullable(String())),
("message", Nullable(String())),
("primary_hash", Nullable(FixedString(32))),
("received", Nullable(DateTime())),
("search_message", Nullable(String())),
("title", Nullable(String())),
("location", Nullable(String())),
# optional user
("user_id", Nullable(String())),
("username", Nullable(String())),
("email", Nullable(String())),
("ip_address", Nullable(String())),
# optional geo
("geo_country_code", Nullable(String())),
("geo_region", Nullable(String())),
("geo_city", Nullable(String())),
("sdk_name", Nullable(String())),
("sdk_version", Nullable(String())),
("type", Nullable(String())),
("version", Nullable(String())),
] + metadata_columns + promoted_context_columns +
promoted_tag_columns + promoted_context_tag_columns + [
# other tags
("tags", Nested([("key", String()), ("value", String())])),
("_tags_flattened", String()),
# other context
("contexts", Nested([("key", String()), ("value", String())])),
# http interface
("http_method", Nullable(String())),
("http_referer", Nullable(String())),
# exception interface
(
"exception_stacks",
Nested([
("type", Nullable(String())),
("value", Nullable(String())),
("mechanism_type", Nullable(String())),
("mechanism_handled", Nullable(UInt(8))),
]),
),
(
"exception_frames",
Nested([
("abs_path", Nullable(String())),
("filename", Nullable(String())),
("package", Nullable(String())),
("module", Nullable(String())),
("function", Nullable(String())),
("in_app", Nullable(UInt(8))),
("colno", Nullable(UInt(32))),
("lineno", Nullable(UInt(32))),
("stack_level", UInt(16)),
]),
),
# These are columns we added later in the life of the (current) production
# database. They don't necessarily belong here in a logical/readability sense
# but they are here to match the order of columns in production becase
# `insert_distributed_sync` is very sensitive to column existence and ordering.
("culprit", Nullable(String())),
("sdk_integrations", Array(String())),
("modules", Nested([("name", String()),
("version", String())])),
])
sample_expr = "cityHash64(toString(event_id))"
schema = ReplacingMergeTreeSchema(
columns=all_columns,
local_table_name="sentry_local",
dist_table_name="sentry_dist",
mandatory_conditions=[("deleted", "=", 0)],
prewhere_candidates=[
"event_id",
"group_id",
"tags[sentry:release]",
"message",
"environment",
"project_id",
],
order_by="(project_id, toStartOfDay(timestamp), %s)" % sample_expr,
partition_by=
"(toMonday(timestamp), if(equals(retention_days, 30), 30, 90))",
version_column="deleted",
sample_expr=sample_expr,
migration_function=events_migrations,
)
dataset_schemas = DatasetSchemas(
read_schema=schema,
write_schema=schema,
)
table_writer = TableWriter(
write_schema=schema,
stream_loader=KafkaStreamLoader(
processor=EventsProcessor(promoted_tag_columns),
default_topic="events",
replacement_topic="event-replacements",
commit_log_topic="snuba-commit-log",
),
)
super(EventsDataset, self).__init__(
dataset_schemas=dataset_schemas,
table_writer=table_writer,
time_group_columns={
"time": "timestamp",
"rtime": "received"
},
time_parse_columns=("timestamp", "received"),
)
self.__metadata_columns = metadata_columns
self.__promoted_tag_columns = promoted_tag_columns
self.__promoted_context_tag_columns = promoted_context_tag_columns
self.__promoted_context_columns = promoted_context_columns
self.__required_columns = required_columns
self.__tags_processor = TagColumnProcessor(
columns=all_columns,
promoted_columns=self._get_promoted_columns(),
column_tag_map=self._get_column_tag_map(),
)
Nullable,
String,
UInt,
UUID,
WithCodecs,
WithDefault,
)
from snuba.migrations.parse_schema import _get_column
test_data = [
# Basic types
(("Date", "", "", ""), Date()),
(("DateTime", "", "", ""), DateTime()),
(("Enum8('success' = 0, 'error' = 1)", "", "", ""),
Enum([("success", 0), ("error", 1)])),
(("FixedString(32)", "", "", ""), FixedString(32)),
(("Float32", "", "", ""), Float(32)),
(("IPv4", "", "", ""), IPv4()),
(("IPv6", "", "", ""), IPv6()),
(("String", "", "", ""), String()),
(("UInt32", "", "", ""), UInt(32)),
(("UUID", "", "", ""), UUID()),
# Aggregate functions
(("AggregateFunction(uniq, UInt8)", "", "", ""),
AggregateFunction("uniq", UInt(8))),
(("AggregateFunction(countIf, UUID, UInt8)", "", "", ""),
AggregateFunction("countIf", UUID(), UInt(8))),
(("AggregateFunction(quantileIf(0.5, 0.9), UInt32, UInt8)", "", "", ""),
AggregateFunction("quantileIf(0.5, 0.9)", UInt(32), UInt(8))),
# Array
(("Array(String)", "", "", ""), Array(String())),
IPv6,
String,
UInt,
)
from snuba.migrations.columns import MigrationModifiers as Modifiers
from snuba.migrations.parse_schema import _get_column
test_data = [
# Basic types
(("Date", "", "", ""), Date()),
(("DateTime", "", "", ""), DateTime()),
(
("Enum8('success' = 0, 'error' = 1)", "", "", ""),
Enum([("success", 0), ("error", 1)]),
),
(("FixedString(32)", "", "", ""), FixedString(32)),
(("Float32", "", "", ""), Float(32)),
(("IPv4", "", "", ""), IPv4()),
(("IPv6", "", "", ""), IPv6()),
(("String", "", "", ""), String()),
(("UInt32", "", "", ""), UInt(32)),
(("UUID", "", "", ""), UUID()),
# Aggregate functions
(
("AggregateFunction(uniq, UInt8)", "", "", ""),
AggregateFunction("uniq", [UInt(8)]),
),
(
("AggregateFunction(countIf, UUID, UInt8)", "", "", ""),
AggregateFunction("countIf", [UUID(), UInt(8)]),
),
def __init__(self) -> None:
self.__common_columns = ColumnSet([
("event_id", FixedString(32)),
("project_id", UInt(64)),
("type", String(Modifiers(nullable=True))),
("timestamp", DateTime()),
("platform", String(Modifiers(nullable=True))),
("environment", String(Modifiers(nullable=True))),
("release", String(Modifiers(nullable=True))),
("dist", String(Modifiers(nullable=True))),
("user", String(Modifiers(nullable=True))),
("transaction", String(Modifiers(nullable=True))),
("message", String(Modifiers(nullable=True))),
("title", String(Modifiers(nullable=True))),
# User
("user_id", String(Modifiers(nullable=True))),
("username", String(Modifiers(nullable=True))),
("email", String(Modifiers(nullable=True))),
("ip_address", String(Modifiers(nullable=True))),
# SDK
("sdk_name", String(Modifiers(nullable=True))),
("sdk_version", String(Modifiers(nullable=True))),
# geo location context
("geo_country_code", String(Modifiers(nullable=True))),
("geo_region", String(Modifiers(nullable=True))),
("geo_city", String(Modifiers(nullable=True))),
("http_method", String(Modifiers(nullable=True))),
("http_referer", String(Modifiers(nullable=True))),
# Other tags and context
("tags", Nested([("key", String()), ("value", String())])),
("contexts", Nested([("key", String()), ("value", String())])),
("trace_id", String(Modifiers(nullable=True))),
("span_id", UInt(64, Modifiers(nullable=True))),
])
self.__events_columns = EVENTS_COLUMNS
self.__transactions_columns = TRANSACTIONS_COLUMNS
discover_storage = get_storage(StorageKey.DISCOVER)
discover_storage_plan_builder = SingleStorageQueryPlanBuilder(
storage=discover_storage,
mappers=events_translation_mappers.
concat(transaction_translation_mappers).concat(
null_function_translation_mappers).concat(
TranslationMappers(columns=[
ColumnToFunction(
None,
"ip_address",
"coalesce",
(
FunctionCall(
None,
"IPv4NumToString",
(Column(None, None, "ip_address_v4"), ),
),
FunctionCall(
None,
"IPv6NumToString",
(Column(None, None, "ip_address_v6"), ),
),
),
),
ColumnToColumn(None, "transaction", None,
"transaction_name"),
ColumnToColumn(None, "username", None, "user_name"),
ColumnToColumn(None, "email", None, "user_email"),
ColumnToMapping(
None,
"geo_country_code",
None,
"contexts",
"geo.country_code",
nullable=True,
),
ColumnToMapping(
None,
"geo_region",
None,
"contexts",
"geo.region",
nullable=True,
),
ColumnToMapping(
None,
"geo_city",
None,
"contexts",
"geo.city",
nullable=True,
),
ColumnToFunction(
None,
"user",
"nullIf",
(Column(None, None, "user"), Literal(None, "")),
),
])).concat(
TranslationMappers(subscriptables=[
SubscriptableMapper(None, "tags", None, "tags"),
SubscriptableMapper(None, "contexts", None,
"contexts"),
], )),
)
discover_pipeline_builder = SimplePipelineBuilder(
query_plan_builder=discover_storage_plan_builder)
super().__init__(
storages=[discover_storage],
query_pipeline_builder=discover_pipeline_builder,
abstract_column_set=(self.__common_columns +
self.__events_columns +
self.__transactions_columns),
join_relationships={},
writable_storage=None,
validators=[EntityRequiredColumnValidator({"project_id"})],
required_time_column="timestamp",
)
def __init__(self) -> None:
self.__common_columns = ColumnSet([
("event_id", FixedString(32)),
("project_id", UInt(64)),
("type", String(Modifiers(nullable=True))),
("timestamp", DateTime()),
("platform", String(Modifiers(nullable=True))),
("environment", String(Modifiers(nullable=True))),
("release", String(Modifiers(nullable=True))),
("dist", String(Modifiers(nullable=True))),
("user", String(Modifiers(nullable=True))),
("transaction", String(Modifiers(nullable=True))),
("message", String(Modifiers(nullable=True))),
("title", String(Modifiers(nullable=True))),
# User
("user_id", String(Modifiers(nullable=True))),
("username", String(Modifiers(nullable=True))),
("email", String(Modifiers(nullable=True))),
("ip_address", String(Modifiers(nullable=True))),
# SDK
("sdk_name", String(Modifiers(nullable=True))),
("sdk_version", String(Modifiers(nullable=True))),
# geo location context
("geo_country_code", String(Modifiers(nullable=True))),
("geo_region", String(Modifiers(nullable=True))),
("geo_city", String(Modifiers(nullable=True))),
("http_method", String(Modifiers(nullable=True))),
("http_referer", String(Modifiers(nullable=True))),
# Other tags and context
("tags", Nested([("key", String()), ("value", String())])),
("contexts", Nested([("key", String()), ("value", String())])),
])
self.__events_columns = EVENTS_COLUMNS
self.__transactions_columns = TRANSACTIONS_COLUMNS
events_storage = get_storage(StorageKey.EVENTS)
events_pipeline_builder = SimplePipelineBuilder(
query_plan_builder=SelectedStorageQueryPlanBuilder(
selector=EventsQueryStorageSelector(
mappers=events_translation_mappers.
concat(transaction_translation_mappers).concat(
null_function_translation_mappers).concat(
TranslationMappers(
# XXX: Remove once we are using errors
columns=[
ColumnToMapping(None, "release", None,
"tags", "sentry:release"),
ColumnToMapping(None, "dist", None, "tags",
"sentry:dist"),
ColumnToMapping(None, "user", None, "tags",
"sentry:user"),
],
subscriptables=[
SubscriptableMapper(
None, "tags", None, "tags"),
SubscriptableMapper(
None, "contexts", None, "contexts"),
],
)))), )
discover_storage = get_storage(StorageKey.DISCOVER)
discover_pipeline_builder = SimplePipelineBuilder(
query_plan_builder=SingleStorageQueryPlanBuilder(
storage=discover_storage,
mappers=events_translation_mappers.concat(
transaction_translation_mappers).
concat(null_function_translation_mappers).concat(
TranslationMappers(columns=[
ColumnToFunction(
None,
"ip_address",
"coalesce",
(
FunctionCall(
None,
"IPv4NumToString",
(Column(None, None, "ip_address_v4"), ),
),
FunctionCall(
None,
"IPv6NumToString",
(Column(None, None, "ip_address_v6"), ),
),
),
),
ColumnToColumn(None, "transaction", None,
"transaction_name"),
ColumnToColumn(None, "username", None, "user_name"),
ColumnToColumn(None, "email", None, "user_email"),
ColumnToMapping(
None,
"geo_country_code",
None,
"contexts",
"geo.country_code",
),
ColumnToMapping(None, "geo_region", None, "contexts",
"geo.region"),
ColumnToMapping(None, "geo_city", None, "contexts",
"geo.city"),
ColumnToFunction(
None,
"user",
"nullIf",
(Column(None, None, "user"), Literal(None, "")),
),
])).concat(
TranslationMappers(subscriptables=[
SubscriptableMapper(None, "tags", None, "tags"),
SubscriptableMapper(None, "contexts", None,
"contexts"),
], )),
))
def selector_func(_query: Query) -> Tuple[str, List[str]]:
if random.random() < float(
state.get_config("discover_query_percentage", 0)):
return "events", ["discover"]
return "events", []
super().__init__(
storages=[events_storage, discover_storage],
query_pipeline_builder=PipelineDelegator(
query_pipeline_builders={
"events": events_pipeline_builder,
"discover": discover_pipeline_builder,
},
selector_func=selector_func,
callback_func=partial(callback_func, "discover"),
),
abstract_column_set=(self.__common_columns +
self.__events_columns +
self.__transactions_columns),
join_relationships={},
writable_storage=None,
)
pytest.param(
IPv4(None),
IPv4(),
IPv4(Modifier(nullable=True)),
"IPv4",
id="IPs",
),
pytest.param(
IPv6(None),
IPv6(),
IPv6(Modifier(nullable=True)),
"IPv6",
id="IPs",
),
pytest.param(
FixedString(32, Modifier(nullable=True)),
FixedString(32),
FixedString(64, Modifier(nullable=True)),
"Nullable(FixedString(32))",
id="fixed strings",
),
pytest.param(
UInt(8, Modifier(nullable=True)),
UInt(8),
UInt(16, Modifier(nullable=True)),
"Nullable(UInt8)",
id="integers",
),
pytest.param(
Float(64, Modifier(nullable=True)),
Float(64),
def __init__(self) -> None:
self.__common_columns = ColumnSet([
("event_id", FixedString(32)),
("project_id", UInt(64)),
("type", String(Modifiers(nullable=True))),
("timestamp", DateTime()),
("platform", String(Modifiers(nullable=True))),
("environment", String(Modifiers(nullable=True))),
("release", String(Modifiers(nullable=True))),
("dist", String(Modifiers(nullable=True))),
("user", String(Modifiers(nullable=True))),
("transaction", String(Modifiers(nullable=True))),
("message", String(Modifiers(nullable=True))),
("title", String(Modifiers(nullable=True))),
# User
("user_id", String(Modifiers(nullable=True))),
("username", String(Modifiers(nullable=True))),
("email", String(Modifiers(nullable=True))),
("ip_address", String(Modifiers(nullable=True))),
# SDK
("sdk_name", String(Modifiers(nullable=True))),
("sdk_version", String(Modifiers(nullable=True))),
# geo location context
("geo_country_code", String(Modifiers(nullable=True))),
("geo_region", String(Modifiers(nullable=True))),
("geo_city", String(Modifiers(nullable=True))),
("http_method", String(Modifiers(nullable=True))),
("http_referer", String(Modifiers(nullable=True))),
# Other tags and context
("tags", Nested([("key", String()), ("value", String())])),
("contexts", Nested([("key", String()), ("value", String())])),
])
self.__events_columns = EVENTS_COLUMNS
self.__transactions_columns = TRANSACTIONS_COLUMNS
events_storage = get_storage(StorageKey.EVENTS)
super().__init__(
storages=[events_storage],
query_plan_builder=SelectedStorageQueryPlanBuilder(
selector=EventsQueryStorageSelector(
mappers=events_translation_mappers.
concat(transaction_translation_mappers).concat(
null_function_translation_mappers).concat(
TranslationMappers(
# XXX: Remove once we are using errors
columns=[
ColumnToMapping(None, "release", None,
"tags", "sentry:release"),
ColumnToMapping(None, "dist", None, "tags",
"sentry:dist"),
ColumnToMapping(None, "user", None, "tags",
"sentry:user"),
],
subscriptables=[
SubscriptableMapper(
None, "tags", None, "tags"),
SubscriptableMapper(
None, "contexts", None, "contexts"),
],
)))),
abstract_column_set=(self.__common_columns +
self.__events_columns +
self.__transactions_columns),
writable_storage=None,
)
def __init__(self) -> None:
self.__common_columns = ColumnSet([
("event_id", FixedString(32)),
("project_id", UInt(64)),
("type", String(Modifiers(nullable=True))),
("timestamp", DateTime()),
("platform", String(Modifiers(nullable=True))),
("environment", String(Modifiers(nullable=True))),
("release", String(Modifiers(nullable=True))),
("dist", String(Modifiers(nullable=True))),
("user", String(Modifiers(nullable=True))),
("transaction", String(Modifiers(nullable=True))),
("message", String(Modifiers(nullable=True))),
("title", String(Modifiers(nullable=True))),
# User
("user_id", String(Modifiers(nullable=True))),
("username", String(Modifiers(nullable=True))),
("email", String(Modifiers(nullable=True))),
("ip_address", String(Modifiers(nullable=True))),
# SDK
("sdk_name", String(Modifiers(nullable=True))),
("sdk_version", String(Modifiers(nullable=True))),
# geo location context
("geo_country_code", String(Modifiers(nullable=True))),
("geo_region", String(Modifiers(nullable=True))),
("geo_city", String(Modifiers(nullable=True))),
("http_method", String(Modifiers(nullable=True))),
("http_referer", String(Modifiers(nullable=True))),
# Other tags and context
("tags", Nested([("key", String()), ("value", String())])),
("contexts", Nested([("key", String()), ("value", String())])),
("trace_id", String(Modifiers(nullable=True))),
])
self.__events_columns = EVENTS_COLUMNS
self.__transactions_columns = TRANSACTIONS_COLUMNS
events_storage = get_storage(StorageKey.EVENTS)
events_pipeline_builder = SimplePipelineBuilder(
query_plan_builder=SelectedStorageQueryPlanBuilder(
selector=EventsQueryStorageSelector(
mappers=events_translation_mappers.
concat(transaction_translation_mappers).concat(
null_function_translation_mappers).concat(
TranslationMappers(
# XXX: Remove once we are using errors
columns=[
ColumnToMapping(None, "release", None,
"tags", "sentry:release"),
ColumnToMapping(None, "dist", None, "tags",
"sentry:dist"),
ColumnToMapping(None, "user", None, "tags",
"sentry:user"),
],
subscriptables=[
SubscriptableMapper(
None, "tags", None, "tags"),
SubscriptableMapper(
None, "contexts", None, "contexts"),
],
)))), )
discover_storage = get_storage(StorageKey.DISCOVER)
discover_storage_plan_builder = SingleStorageQueryPlanBuilder(
storage=discover_storage,
mappers=events_translation_mappers.
concat(transaction_translation_mappers).concat(
null_function_translation_mappers).concat(
TranslationMappers(columns=[
ColumnToFunction(
None,
"ip_address",
"coalesce",
(
FunctionCall(
None,
"IPv4NumToString",
(Column(None, None, "ip_address_v4"), ),
),
FunctionCall(
None,
"IPv6NumToString",
(Column(None, None, "ip_address_v6"), ),
),
),
),
ColumnToColumn(None, "transaction", None,
"transaction_name"),
ColumnToColumn(None, "username", None, "user_name"),
ColumnToColumn(None, "email", None, "user_email"),
ColumnToMapping(
None,
"geo_country_code",
None,
"contexts",
"geo.country_code",
nullable=True,
),
ColumnToMapping(
None,
"geo_region",
None,
"contexts",
"geo.region",
nullable=True,
),
ColumnToMapping(
None,
"geo_city",
None,
"contexts",
"geo.city",
nullable=True,
),
ColumnToFunction(
None,
"user",
"nullIf",
(Column(None, None, "user"), Literal(None, "")),
),
])).concat(
TranslationMappers(subscriptables=[
SubscriptableMapper(None, "tags", None, "tags"),
SubscriptableMapper(None, "contexts", None,
"contexts"),
], )),
)
discover_pipeline_builder = SimplePipelineBuilder(
query_plan_builder=discover_storage_plan_builder)
pipeline_builder: Union[PipelineDelegator, SimplePipelineBuilder]
if settings.ERRORS_ROLLOUT_ALL:
storage = discover_storage
sampled_pipeline_builder = SampledSimplePipelineBuilder(
query_plan_builder=discover_storage_plan_builder)
pipeline_builder = PipelineDelegator(
query_pipeline_builders={
"primary": discover_pipeline_builder,
"sampler": sampled_pipeline_builder,
},
selector_func=sampling_selector_func,
callback_func=sampling_callback_func,
)
else:
storage = events_storage
pipeline_builder = events_pipeline_builder
super().__init__(
storages=[storage],
query_pipeline_builder=pipeline_builder,
abstract_column_set=(self.__common_columns +
self.__events_columns +
self.__transactions_columns),
join_relationships={},
writable_storage=None,
validators=[EntityRequiredColumnValidator({"project_id"})],
required_time_column="timestamp",
)
def __init__(self) -> None:
self.__common_columns = ColumnSet([
("event_id", FixedString(32)),
("project_id", UInt(64)),
("type", String(Modifiers(nullable=True))),
("timestamp", DateTime()),
("platform", String(Modifiers(nullable=True))),
("environment", String(Modifiers(nullable=True))),
("release", String(Modifiers(nullable=True))),
("dist", String(Modifiers(nullable=True))),
("user", String(Modifiers(nullable=True))),
("transaction", String(Modifiers(nullable=True))),
("message", String(Modifiers(nullable=True))),
("title", String(Modifiers(nullable=True))),
# User
("user_id", String(Modifiers(nullable=True))),
("username", String(Modifiers(nullable=True))),
("email", String(Modifiers(nullable=True))),
("ip_address", String(Modifiers(nullable=True))),
# SDK
("sdk_name", String(Modifiers(nullable=True))),
("sdk_version", String(Modifiers(nullable=True))),
# geo location context
("geo_country_code", String(Modifiers(nullable=True))),
("geo_region", String(Modifiers(nullable=True))),
("geo_city", String(Modifiers(nullable=True))),
("http_method", String(Modifiers(nullable=True))),
("http_referer", String(Modifiers(nullable=True))),
# Other tags and context
("tags", Nested([("key", String()), ("value", String())])),
("contexts", Nested([("key", String()), ("value", String())])),
])
self.__events_columns = EVENTS_COLUMNS
self.__transactions_columns = TRANSACTIONS_COLUMNS
events_storage = get_storage(StorageKey.EVENTS)
events_pipeline_builder = SimplePipelineBuilder(
query_plan_builder=SelectedStorageQueryPlanBuilder(
selector=EventsQueryStorageSelector(
mappers=events_translation_mappers.
concat(transaction_translation_mappers).concat(
null_function_translation_mappers).concat(
TranslationMappers(
# XXX: Remove once we are using errors
columns=[
ColumnToMapping(None, "release", None,
"tags", "sentry:release"),
ColumnToMapping(None, "dist", None, "tags",
"sentry:dist"),
ColumnToMapping(None, "user", None, "tags",
"sentry:user"),
],
subscriptables=[
SubscriptableMapper(
None, "tags", None, "tags"),
SubscriptableMapper(
None, "contexts", None, "contexts"),
],
)))), )
discover_storage = get_storage(StorageKey.DISCOVER)
discover_pipeline_builder = SimplePipelineBuilder(
query_plan_builder=SingleStorageQueryPlanBuilder(
storage=discover_storage,
mappers=events_translation_mappers.concat(
transaction_translation_mappers).
concat(null_function_translation_mappers).concat(
TranslationMappers(columns=[
ColumnToFunction(
None,
"ip_address",
"coalesce",
(
FunctionCall(
None,
"IPv4NumToString",
(Column(None, None, "ip_address_v4"), ),
),
FunctionCall(
None,
"IPv6NumToString",
(Column(None, None, "ip_address_v6"), ),
),
),
),
ColumnToColumn(None, "transaction", None,
"transaction_name"),
ColumnToColumn(None, "username", None, "user_name"),
ColumnToColumn(None, "email", None, "user_email"),
ColumnToMapping(
None,
"geo_country_code",
None,
"contexts",
"geo.country_code",
nullable=True,
),
ColumnToMapping(
None,
"geo_region",
None,
"contexts",
"geo.region",
nullable=True,
),
ColumnToMapping(
None,
"geo_city",
None,
"contexts",
"geo.city",
nullable=True,
),
ColumnToFunction(
None,
"user",
"nullIf",
(Column(None, None, "user"), Literal(None, "")),
),
])).concat(
TranslationMappers(subscriptables=[
SubscriptableMapper(None, "tags", None, "tags"),
SubscriptableMapper(None, "contexts", None,
"contexts"),
], )),
))
def selector_func(_query: Query,
referrer: str) -> Tuple[str, List[str]]:
# In case something goes wrong, set this to 1 to revert to the events storage.
kill_rollout = state.get_config("errors_rollout_killswitch", 0)
assert isinstance(kill_rollout, (int, str))
if int(kill_rollout):
return "events", []
if referrer in settings.ERRORS_ROLLOUT_BY_REFERRER:
return "discover", []
if settings.ERRORS_ROLLOUT_ALL:
return "discover", []
default_threshold = state.get_config("discover_query_percentage",
0)
assert isinstance(default_threshold, (float, int, str))
threshold = settings.ERRORS_QUERY_PERCENTAGE_BY_REFERRER.get(
referrer, default_threshold)
if random.random() < float(threshold):
return "events", ["discover"]
return "events", []
super().__init__(
storages=[events_storage, discover_storage],
query_pipeline_builder=PipelineDelegator(
query_pipeline_builders={
"events": events_pipeline_builder,
"discover": discover_pipeline_builder,
},
selector_func=selector_func,
callback_func=partial(callback_func, "discover"),
),
abstract_column_set=(self.__common_columns +
self.__events_columns +
self.__transactions_columns),
join_relationships={},
writable_storage=None,
required_filter_columns=["project_id"],
required_time_column="timestamp",
)
("tags", Nested([("key", String()), ("value", String())])),
("_tags_flattened", String()),
("_tags_hash_map", Materialized(Array(UInt(64)), TAGS_HASH_MAP_COLUMN)),
("contexts", Nested([("key", String()), ("value", String())])),
("_contexts_flattened", String()),
("transaction_name", WithDefault(LowCardinality(String()), "''")),
("transaction_hash", Materialized(UInt(64), "cityHash64(transaction_name)"),),
("span_id", Nullable(UInt(64))),
("trace_id", Nullable(UUID())),
("partition", UInt(16)),
("offset", WithCodecs(UInt(64), ["DoubleDelta", "LZ4"])),
("message_timestamp", DateTime()),
("retention_days", UInt(16)),
("deleted", UInt(8)),
("group_id", UInt(64)),
("primary_hash", FixedString(32)),
("primary_hash_hex", Materialized(UInt(64), "hex(primary_hash)")),
("event_string", WithCodecs(String(), ["NONE"])),
("received", DateTime()),
("message", String()),
("title", String()),
("culprit", String()),
("level", LowCardinality(String())),
("location", Nullable(String())),
("version", LowCardinality(Nullable(String()))),
("type", LowCardinality(String())),
(
"exception_stacks",
Nested(
[
("type", Nullable(String())),
def __init__(self):
metadata_columns = ColumnSet([
# optional stream related data
('offset', Nullable(UInt(64))),
('partition', Nullable(UInt(16))),
])
promoted_tag_columns = ColumnSet([
# These are the classic tags, they are saved in Snuba exactly as they
# appear in the event body.
('level', Nullable(String())),
('logger', Nullable(String())),
('server_name', Nullable(String())), # future name: device_id?
('transaction', Nullable(String())),
('environment', Nullable(String())),
('sentry:release', Nullable(String())),
('sentry:dist', Nullable(String())),
('sentry:user', Nullable(String())),
('site', Nullable(String())),
('url', Nullable(String())),
])
promoted_context_tag_columns = ColumnSet([
# These are promoted tags that come in in `tags`, but are more closely
# related to contexts. To avoid naming confusion with Clickhouse nested
# columns, they are stored in the database with s/./_/
# promoted tags
('app_device', Nullable(String())),
('device', Nullable(String())),
('device_family', Nullable(String())),
('runtime', Nullable(String())),
('runtime_name', Nullable(String())),
('browser', Nullable(String())),
('browser_name', Nullable(String())),
('os', Nullable(String())),
('os_name', Nullable(String())),
('os_rooted', Nullable(UInt(8))),
])
promoted_context_columns = ColumnSet([
('os_build', Nullable(String())),
('os_kernel_version', Nullable(String())),
('device_name', Nullable(String())),
('device_brand', Nullable(String())),
('device_locale', Nullable(String())),
('device_uuid', Nullable(String())),
('device_model_id', Nullable(String())),
('device_arch', Nullable(String())),
('device_battery_level', Nullable(Float(32))),
('device_orientation', Nullable(String())),
('device_simulator', Nullable(UInt(8))),
('device_online', Nullable(UInt(8))),
('device_charging', Nullable(UInt(8))),
])
required_columns = ColumnSet([
('event_id', FixedString(32)),
('project_id', UInt(64)),
('group_id', UInt(64)),
('timestamp', DateTime()),
('deleted', UInt(8)),
('retention_days', UInt(16)),
])
all_columns = required_columns + [
# required for non-deleted
('platform', Nullable(String())),
('message', Nullable(String())),
('primary_hash', Nullable(FixedString(32))),
('received', Nullable(DateTime())),
('search_message', Nullable(String())),
('title', Nullable(String())),
('location', Nullable(String())),
# optional user
('user_id', Nullable(String())),
('username', Nullable(String())),
('email', Nullable(String())),
('ip_address', Nullable(String())),
# optional geo
('geo_country_code', Nullable(String())),
('geo_region', Nullable(String())),
('geo_city', Nullable(String())),
('sdk_name', Nullable(String())),
('sdk_version', Nullable(String())),
('type', Nullable(String())),
('version', Nullable(String())),
] + metadata_columns \
+ promoted_context_columns \
+ promoted_tag_columns \
+ promoted_context_tag_columns \
+ [
# other tags
('tags', Nested([
('key', String()),
('value', String()),
])),
# other context
('contexts', Nested([
('key', String()),
('value', String()),
])),
# http interface
('http_method', Nullable(String())),
('http_referer', Nullable(String())),
# exception interface
('exception_stacks', Nested([
('type', Nullable(String())),
('value', Nullable(String())),
('mechanism_type', Nullable(String())),
('mechanism_handled', Nullable(UInt(8))),
])),
('exception_frames', Nested([
('abs_path', Nullable(String())),
('filename', Nullable(String())),
('package', Nullable(String())),
('module', Nullable(String())),
('function', Nullable(String())),
('in_app', Nullable(UInt(8))),
('colno', Nullable(UInt(32))),
('lineno', Nullable(UInt(32))),
('stack_level', UInt(16)),
])),
# These are columns we added later in the life of the (current) production
# database. They don't necessarily belong here in a logical/readability sense
# but they are here to match the order of columns in production becase
# `insert_distributed_sync` is very sensitive to column existence and ordering.
('culprit', Nullable(String())),
('sdk_integrations', Array(String())),
('modules', Nested([
('name', String()),
('version', String()),
])),
]
sample_expr = 'cityHash64(toString(event_id))'
schema = ReplacingMergeTreeSchema(
columns=all_columns,
local_table_name='sentry_local',
dist_table_name='sentry_dist',
mandatory_conditions=[('deleted', '=', 0)],
order_by='(project_id, toStartOfDay(timestamp), %s)' % sample_expr,
partition_by='(toMonday(timestamp), if(equals(retention_days, 30), 30, 90))',
version_column='deleted',
sample_expr=sample_expr,
migration_function=events_migrations)
dataset_schemas = DatasetSchemas(
read_schema=schema,
write_schema=schema,
)
table_writer = TableWriter(
write_schema=schema,
stream_loader=KafkaStreamLoader(
processor=EventsProcessor(promoted_tag_columns),
default_topic="events",
replacement_topic="event-replacements",
commit_log_topic="snuba-commit-log",
)
)
super(EventsDataset, self).__init__(
dataset_schemas=dataset_schemas,
table_writer=table_writer,
time_group_columns={
'time': 'timestamp',
'rtime': 'received'
},
time_parse_columns=('timestamp', 'received')
)
self.__metadata_columns = metadata_columns
self.__promoted_tag_columns = promoted_tag_columns
self.__promoted_context_tag_columns = promoted_context_tag_columns
self.__promoted_context_columns = promoted_context_columns
self.__required_columns = required_columns
self.__tags_processor = TagColumnProcessor(
columns=all_columns,
promoted_columns=self._get_promoted_columns(),
column_tag_map=self._get_column_tag_map(),
)
def visit_fixedstring(
self, node: Node, visited_children: Iterable[Any]
) -> ColumnType:
size = int(node.children[3].text)
return FixedString(size)
def __init__(self) -> None:
all_columns = ColumnSet([
("org_id", UInt(64)),
("project_id", UInt(64)),
("timestamp", DateTime()),
("event_id", WithCodecs(UUID(), ["NONE"])),
(
"event_hash",
WithCodecs(
Materialized(
UInt(64),
"cityHash64(toString(event_id))",
),
["NONE"],
),
),
("platform", LowCardinality(String())),
("environment", LowCardinality(Nullable(String()))),
("release", LowCardinality(Nullable(String()))),
("dist", LowCardinality(Nullable(String()))),
("ip_address_v4", Nullable(IPv4())),
("ip_address_v6", Nullable(IPv6())),
("user", WithDefault(String(), "''")),
(
"user_hash",
Materialized(UInt(64), "cityHash64(user)"),
),
("user_id", Nullable(String())),
("user_name", Nullable(String())),
("user_email", Nullable(String())),
("sdk_name", LowCardinality(Nullable(String()))),
("sdk_version", LowCardinality(Nullable(String()))),
("tags", Nested([("key", String()), ("value", String())])),
("_tags_flattened", String()),
("contexts", Nested([("key", String()), ("value", String())])),
("_contexts_flattened", String()),
("transaction_name", WithDefault(LowCardinality(String()), "''")),
(
"transaction_hash",
Materialized(UInt(64), "cityHash64(transaction_name)"),
),
("span_id", Nullable(UInt(64))),
("trace_id", Nullable(UUID())),
("partition", UInt(16)),
("offset", WithCodecs(UInt(64), ["DoubleDelta", "LZ4"])),
("retention_days", UInt(16)),
("deleted", UInt(8)),
("group_id", UInt(64)),
("primary_hash", FixedString(32)),
("primary_hash_hex", Materialized(UInt(64), "hex(primary_hash)")),
("event_string", WithCodecs(String(), ["NONE"])),
("received", DateTime()),
("message", String()),
("title", String()),
("culprit", String()),
("level", LowCardinality(String())),
("location", Nullable(String())),
("version", LowCardinality(Nullable(String()))),
("type", LowCardinality(String())),
(
"exception_stacks",
Nested([
("type", Nullable(String())),
("value", Nullable(String())),
("mechanism_type", Nullable(String())),
("mechanism_handled", Nullable(UInt(8))),
]),
),
(
"exception_frames",
Nested([
("abs_path", Nullable(String())),
("colno", Nullable(UInt(32))),
("filename", Nullable(String())),
("function", Nullable(String())),
("lineno", Nullable(UInt(32))),
("in_app", Nullable(UInt(8))),
("package", Nullable(String())),
("module", Nullable(String())),
("stack_level", Nullable(UInt(16))),
]),
),
("sdk_integrations", Array(String())),
("modules", Nested([("name", String()), ("version", String())])),
])
self.__promoted_tag_columns = {
"environment": "environment",
"sentry:release": "release",
"sentry:dist": "dist",
"sentry:user": "******",
"transaction": "transaction_name",
"level": "level",
}
schema = ReplacingMergeTreeSchema(
columns=all_columns,
local_table_name="errors_local",
dist_table_name="errors_dist",
mandatory_conditions=[("deleted", "=", 0)],
prewhere_candidates=[
"event_id",
"group_id",
"tags[sentry:release]",
"message",
"environment",
"project_id",
],
order_by=
"(org_id, project_id, toStartOfDay(timestamp), primary_hash_hex, event_hash)",
partition_by=
"(toMonday(timestamp), if(retention_days = 30, 30, 90))",
version_column="deleted",
sample_expr="event_hash",
ttl_expr="timestamp + toIntervalDay(retention_days)",
settings={"index_granularity": "8192"},
)
dataset_schemas = DatasetSchemas(
read_schema=schema,
write_schema=schema,
)
table_writer = TableWriter(
write_schema=schema,
stream_loader=KafkaStreamLoader(
processor=ErrorsProcessor(self.__promoted_tag_columns),
default_topic="events",
),
)
super().__init__(
dataset_schemas=dataset_schemas,
table_writer=table_writer,
time_group_columns={
"time": "timestamp",
"rtime": "received"
},
time_parse_columns=("timestamp", "received"),
)
self.__tags_processor = TagColumnProcessor(
columns=all_columns,
promoted_columns=self._get_promoted_columns(),
column_tag_map=self._get_column_tag_map(),
)