[
https://issues.apache.org/jira/browse/AURORA-134?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Bill Farner reopened AURORA-134:
--------------------------------
> LogStorage store implementations internally invoke write
> --------------------------------------------------------
>
> Key: AURORA-134
> URL: https://issues.apache.org/jira/browse/AURORA-134
> Project: Aurora
> Issue Type: Bug
> Components: Scheduler
> Reporter: Bill Farner
> Assignee: Bill Farner
> Priority: Minor
>
> {{LogStorage}} implements all of the *Store interfaces, and is the outer-most
> implementation provided to callers throughout the scheduler (it delegates to
> MemStorage). We recently noticed that {{LogStorage}} implements all mutating
> store methods by immediately invoking write()., which is odd since external
> callers should only gain access to these methods by first invoking write().
> This has the outcome of an additional write() call for all mutate operations.
> For example:
> {code}
> storage.write(new MutateWork.NoResult.Quiet() {
> @Override protected void execute(MutableStoreProvider storeProvider) {
> for (xx) {
> storeProvider.getAttributeStore().saveHostAttributes(..);
> }
> }
> });
> {code}
> If the inner {{for}} loop executes N times, this block of code ends up
> invoking LogStorage.write N+1 times because of this behavior.
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
