You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Commit 508709b
Browse filesBrowse the repository at this point in the historyBrowse files
perf(kernel): skip the Python telemetry client on use_kernel connections
The kernel exports telemetry for use_kernel connections, but every
kernel connect still registered a no-op client with
TelemetryClientFactory under its global lock (and the first one started
the flush thread). In a 200-thread cold burst that lock convoys behind
the GIL: initialize_telemetry_client took p50 692 ms / max 1.7 s, and
connect p50 was ~3.0 s vs ~1.7 s without it.
Kernel connections now use a NoopTelemetryClient directly and skip the
factory on close. This also stops a kernel connection opened first from
leaving a no-op client registered for the host, which silently disabled
Python telemetry for later Thrift connections to that host.
Co-authored-by: Isaac <no-reply@databricks.com>
Signed-off-by: Vu Anh Phung <vu.phung@databricks.com>
Copy file name to clipboardExpand all lines: CHANGELOG.md
+1Lines changed: 1 addition & 0 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -4,6 +4,7 @@
4
4
- Fix the kernel backend ignoring `use_cloud_fetch=False`, so complete results can be fetched without cloud-storage downloads.
5
5
- Fix inline parameters (`use_inline_params=True`) so a sequence directly after `IN` renders as a list again: `x IN %(ids)s` renders as `x IN (1,2,3)` instead of `x IN ARRAY(1,2,3)`, which Spark rejects. Inside the list, tuples render as rows for a multi-column `IN`, and a `set` or `frozenset` is accepted again. Sequences anywhere else still render as `ARRAY(...)` ([#629](https://github.com/databricks/databricks-sql-python/issues/629))
6
6
- Speed up Thrift response handling: read replies in 1 MiB chunks and decode them with Thrift's C-accelerated protocol, and stop stringifying every request/response for the default no-op debug logger. Wide result schemas, metadata calls with many rows, and inline Arrow results benefit most.
7
+
- Speed up concurrent `use_kernel=True` connects by skipping the Python telemetry client setup, which the kernel does not use; this also stops a kernel connection from disabling Python telemetry for later Thrift connections to the same host.
7
8
8
9
# 4.9.0 (2026-11-04)
9
10
- Upgrade Databricks SQL Kernel to 1.4.0, which keeps Lakehouse Real-Time session affinity and flushes shared telemetry only after the final session closes.
0 commit comments