PYTHON-4227 Unified tests: Advance cluster_time of ClientSessions after initialData creation #1603
Conversation
|
I'm investigating the new failure that popped up. I can reproduce locally. |
ShaneHarvey
changed the title
test/unified_format.py
Outdated
| if cluster_time: | ||
| for entity in self.entity_map._entities.values(): | ||
| if isinstance(entity, ClientSession): | ||
| entity.advance_cluster_time(cluster_time) |
There was a problem hiding this comment.
We need to do this for sessions created later on in _testOperation_createEntities too.
There was a problem hiding this comment.
Where is _testOperation_createEntities called from? Would we want to ping again before updating?
There was a problem hiding this comment.
We should only run ping once, store the cluster_time, then use it later when _testOperation_createEntities is called.
test/unified_format.py
Outdated
| # to ensure consistency in transactions against a sharded deployment, | ||
| cluster_time = self.client.admin.command("ping").get("$clusterTime") | ||
| if cluster_time: | ||
| for entity in self.entity_map._entities.values(): |
There was a problem hiding this comment.
Suggest adding a EntityMapUtil.values() method that we can call here so we don't need to look at the internals.
|
@ShaneHarvey, @blink1073 |
Yeah, apparently https://jira.mongodb.org/browse/PYTHON-4182 was unrelated, I unblocked it and put it in scheduled. |
test/unified_format.py
Outdated
| """Manually synchronize entities when desired""" | ||
| if not self._cluster_time: | ||
| self._cluster_time = self.test.client.admin.command("ping").get("$clusterTime") | ||
| for entity in self._entities: |
There was a problem hiding this comment.
| for entity in self._entities: | |
| for entity in self._entities.values(): |
The current version iterates over the names instead of the values and never passes the isinstance check.
test/unified_format.py
Outdated
| if "unpin after non-transient error on abort" in spec["description"]: | ||
| if client_context.version[0] == 8: | ||
| self.skipTest("Skipping TransientTransactionError pending PYTHON-4182") | ||
| if "unpin after TransientTransactionError error on" in spec["description"]: |
There was a problem hiding this comment.
Is this intended? The failure is the same for PYTHON-4182, so I would expect this PR to fix that ticket as well.
There was a problem hiding this comment.
I can try it out. The failure was popping back up earlier but now that everything is cleaner, and I've synced everything up to master, it is worth trying again.
…ending PYTHON-4182
caseyclements
force-pushed
the
PYTHON-4227-advance-clustertime-in-unified-tests
branch
from
2afe41b
to
8bc865b
Compare
|
Merging so we can check the waterfall afterward. |
Changes
After initialData is added in test.unified_format's run_scenario, we get the $clusterTime from the server, and add this to each of the ClientSession entities before further operations.
Motivation for the change.
DRIVERS-2816
Here is the pertinent section of the specifications