Details
-
Bug
-
Resolution: Duplicate
-
Major
-
None
-
None
-
None
-
None
-
0
Description
To reproduce
from datetime import timedelta |
from couchbase.cluster import Cluster |
from couchbase.options import ClusterOptions, ClusterTimeoutOptions |
from couchbase.auth import PasswordAuthenticator |
|
|
opts = ClusterOptions(PasswordAuthenticator('Administrator', 'password'), timeout_options=ClusterTimeoutOptions(analytics_timeout=timedelta(seconds=5))) |
cluster = Cluster('couchbase://localhost', opts) |
result = cluster.analytics_query('select value sleep("result", 100000)') |
for row in result: |
print(f"Row: {row}") |
This should have timed out at 5 seconds. It does not, it times out after 75 seconds which is the default analytics timeout, so it looks like the analytics timeout set through the ClusterTimeoutOptions is not respected. From an initial look, it seems that something similar applies to all timeouts in ClusterTimeoutOptions.
The cause appears to be that even though the timeout options are appended to the cluster options and the 'timeout_options' key is removed (see 1), on the C++ side we are trying to find the 'timeout_options' key (see 2).
(1) couchbase/logic/options.py
timeout_opts = kwargs.pop('timeout_options', {}) |
if timeout_opts: |
for k, v in timeout_opts.items(): |
if k not in kwargs: |
kwargs[k] = v |
(2) src/connection.cxx
PyObject* pyObj_timeout_opts = PyDict_GetItemString(pyObj_options, "timeout_options"); |
if (pyObj_timeout_opts != nullptr) { |
update_cluster_timeout_options(options, pyObj_timeout_opts);
|
}
|
Attachments
Issue Links
- duplicates
-
-
- Closed
-