wassname/ray
1
0
Fork 0
mirror of https://github.com/wassname/ray.git synced 2026-07-03 03:10:54 +08:00
Code Issues Packages Projects Releases Wiki Activity

Print error when actor takes too long to start, and refactor error me… (#1747)

Browse Source 
* Print error when actor takes too long to start, and refactor error message pushing.

* Print warning every ten seconds.

* Fix linting and tests.

* Fix tests.
This commit is contained in:
Robert Nishihara
2018-03-19 20:24:35 -07:00
committed by Philipp Moritz
parent 73bb149c8a
commit 4658d0a180
10 changed files with 103 additions and 90 deletions
Download Patch File Download Diff File Expand all files Collapse all files
python/ray/worker.py
-17
View File
@@ -1100,23 +1100,6 @@ def error_info(worker=global_worker):
for error_key in error_keys:
if error_applies_to_driver(error_key, worker=worker):
error_contents = worker.redis_client.hgetall(error_key)
# If the error is an object hash mismatch, look up the function
# name for the nondeterministic task. TODO(rkn): Change this so
# that we don't have to look up additional information. Ideally all
# relevant information would already be in error_contents.
error_type = error_contents[b"type"]
if error_type in [OBJECT_HASH_MISMATCH_ERROR_TYPE,
PUT_RECONSTRUCTION_ERROR_TYPE]:
function_id = error_contents[b"data"]
if function_id == NIL_FUNCTION_ID:
function_name = b"Driver"
else:
task_driver_id = worker.task_driver_id
function_name = worker.redis_client.hget(
(b"RemoteFunction:" + task_driver_id.id() +
b":" + function_id),
"name")
error_contents[b"data"] = function_name
errors.append(error_contents)
return errors