We've had some back end db issues since switching to unified login. I was looking at some logs today as Teaneck is currently down. I'm seeing a lot of errors about not being able to find free/available db connections.
One small thing I'm trying here is to change a bunch of instances where we're using withTransaction and switching it to withSession. I think that in some cases this may result in essentially a lock being taken out on a table, preventing it from being used by other processes, which could be slowing down the system.
It looked like Teaneck may have gone down while running nightly clustering, and I took a quick peek at the code and noticed that we were using withTransaction often there unnecessarily, which is why I started making this change. I'm honestly not expecting it to really help, but you never know! And it's something that we've needed to do anyway.
Other small updates:
Updated instance of .length.run to .size.run. Both of these get a count of rows in the db, but the former will be deprecated in the future, and we should keep things consistent!
A few functions were simplified, mostly for clarity. Primarily in DBTableDefinitions.scala and SurveyQuestionTable.scala
Fixed a couple instances of nested transactions (#3550) in LabelTable.scala and MissionTable.scala
Things to check before submitting the PR
[x] I've written a descriptive PR title.
[x] I've added/updated comments for large or confusing blocks of code.
Working on #3726
We've had some back end db issues since switching to unified login. I was looking at some logs today as Teaneck is currently down. I'm seeing a lot of errors about not being able to find free/available db connections.
One small thing I'm trying here is to change a bunch of instances where we're using
withTransactionand switching it towithSession. I think that in some cases this may result in essentially a lock being taken out on a table, preventing it from being used by other processes, which could be slowing down the system.It looked like Teaneck may have gone down while running nightly clustering, and I took a quick peek at the code and noticed that we were using
withTransactionoften there unnecessarily, which is why I started making this change. I'm honestly not expecting it to really help, but you never know! And it's something that we've needed to do anyway.Other small updates:
.length.runto.size.run. Both of these get a count of rows in the db, but the former will be deprecated in the future, and we should keep things consistent!DBTableDefinitions.scalaandSurveyQuestionTable.scalaLabelTable.scalaandMissionTable.scalaThings to check before submitting the PR