def update_cached_items() -> None:
tasks = []
items = (Insight.objects.filter(
Q(
Q(dashboard__is_shared=True)
| Q(dashboard__last_accessed_at__gt=timezone.now() -
relativedelta(days=7)))).exclude(
dashboard__deleted=True).exclude(refreshing=True).exclude(
deleted=True).exclude(refresh_attempt__gt=2).exclude(
filters={}).order_by(
F("last_refresh").asc(nulls_first=True)))
for item in items[0:PARALLEL_INSIGHT_CACHE]:
try:
cache_key, cache_type, payload = dashboard_item_update_task_params(
item)
if item.filters_hash != cache_key:
item.save(
) # force update if the saved key is different from the cache key
tasks.append(
update_cache_item_task.s(cache_key, cache_type, payload))
except Exception as e:
item.refresh_attempt = (item.refresh_attempt or 0) + 1
item.save()
capture_exception(e)
logger.info("Found {} items to refresh".format(len(tasks)))
taskset = group(tasks)
taskset.apply_async()
statsd.gauge("update_cache_queue_depth", items.count())
def redis_celery_queue_depth():
from statshog.defaults.django import statsd
try:
llen = get_client().llen("celery")
statsd.gauge(f"posthog_celery_queue_depth", llen)
except:
# if we can't connect to statsd don't complain about it.
# not every installation will have statsd available
return
def sync_execute(query, args=None, settings=None):
with ch_pool.get_client() as client:
start_time = time()
tags = {}
try:
sql, tags = _annotate_tagged_query(query, args)
result = client.execute(sql, args, settings=settings)
finally:
execution_time = time() - start_time
statsd.gauge("clickhouse_sync_execution_time", execution_time * 1000.0, tags=tags)
if app_settings.SHELL_PLUS_PRINT_SQL:
print(format_sql(query, args))
print("Execution time: %.6fs" % (execution_time,))
if _request_information is not None and _request_information.get("save", False):
save_query(query, args, execution_time)
return result
def clickhouse_row_count():
if is_ee_enabled() and settings.EE_AVAILABLE:
from statshog.defaults.django import statsd
from ee.clickhouse.client import sync_execute
for table in CLICKHOUSE_TABLES:
try:
QUERY = """select count(1) freq from {table};"""
query = QUERY.format(table=table)
rows = sync_execute(query)[0][0]
statsd.gauge(f"posthog_celery_clickhouse_table_row_count",
rows,
tags={"table": table})
except:
pass
else:
pass
def clickhouse_part_count():
if is_ee_enabled() and settings.EE_AVAILABLE:
from statshog.defaults.django import statsd
from ee.clickhouse.client import sync_execute
QUERY = """
select table, count(1) freq
from system.parts
group by table
order by freq desc;
"""
rows = sync_execute(QUERY)
for (table, parts) in rows:
statsd.gauge(f"posthog_celery_clickhouse_table_parts_count",
parts,
tags={"table": table})
else:
pass
def clickhouse_lag():
if is_ee_enabled() and settings.EE_AVAILABLE:
from statshog.defaults.django import statsd
from ee.clickhouse.client import sync_execute
for table in CLICKHOUSE_TABLES:
try:
QUERY = (
"""select max(_timestamp) observed_ts, now() now_ts, now() - max(_timestamp) as lag from {table};"""
)
query = QUERY.format(table=table)
lag = sync_execute(query)[0][2]
statsd.gauge("posthog_celery_clickhouse__table_lag_seconds",
lag,
tags={"table": table})
except:
pass
else:
pass
def clickhouse_mutation_count():
if is_ee_enabled() and settings.EE_AVAILABLE:
from statshog.defaults.django import statsd
from ee.clickhouse.client import sync_execute
QUERY = """
SELECT
table,
count(1) AS freq
FROM system.mutations
GROUP BY table
ORDER BY freq DESC
"""
rows = sync_execute(QUERY)
for (table, muts) in rows:
statsd.gauge(f"posthog_celery_clickhouse_table_mutations_count",
muts,
tags={"table": table})
else:
pass
def gauge(metric_name: str, value: Union[int, float], tags: Tags = None):
statsd.gauge(metric_name, value, tags=tags)
_capture(metric_name, value, tags)
