Cleaning up potential resource handling errors.#1258
Conversation
…clob in try-with-resources blocks. Calling jOOQ fetchStream in try-with-resources.
|
@MikeNeilson Not immediately obvious to me why this build is failing |
https://github.com/USACE/cwms-data-api/pull/1258/files#annotation_38309067931 Granted that still doesn't make it obvious. All of those errors should be in a log in the build directory, we should probably upload it as an action artifact that so we can review it. |
… consumes the streams synchronously.
|
Its hit-or-miss for me whether that annotations link shows anything of not. The github ui seems to be inconsistent. |
|
I've added to the pr and I believe it will fix #1252 |
…nd in: https://tomcat.apache.org/tomcat-9.0-doc/jdbc-pool.html I think they are meant to be removeAbandoned.
Yeah, it tends not to show on the file unless it's a file that was edited and shows in the PR. |
|
It's failing on the Time Series integration test that inserts a massive amount of data. |
|
Something has changed in the backend. I ran the timeseries test_create_big that inserts 200,000 points and the connection gets forceably closed in 45 seconds and the insert still hasn't completed. I'm glad the test is failing if its taking that long but the test used to pass. Is it the change that we are now actually closing the connections? Dialing down the number of points to 10,000 and the zstore_ts method is still taking over 30s to complete. And that is after I moved the json serialization (400ms) and ZTSV array creation (30ms) so it happens before we get a connection to do the zstore_ts call. Maybe I create a new issue to figure out why the timeseries store is so slow? |
Yeah, the only recent change I can think of was for retrieve data entry, at least within CDA, but it's definitely possible a database change screwed something up. |
Explicitly calling blob and clob free. Putting streams from blob and clob in try-with-resources blocks. Calling jOOQ fetchStream in try-with-resources. I don't think any of these is actually causing connection leaks but they seem like easy changes for better safety (cherry picked from commit a660ea5)
Explicitly calling blob and clob free. Putting streams from blob and clob in try-with-resources blocks. Calling jOOQ fetchStream in try-with-resources. I don't think any of these is actually causing connection leaks but they seem like easy changes for better safety (cherry picked from commit a660ea5)
2 participants
Explicitly calling blob and clob free. Putting streams from blob and clob in try-with-resources blocks. Calling jOOQ fetchStream in try-with-resources.
I don't think any of these is actually causing connection leaks but they seem like easy changes for better safety