ibessonov commented on code in PR #7917:
URL: https://github.com/apache/ignite-3/pull/7917#discussion_r3071372643
##########
modules/page-memory/src/integrationTest/java/org/apache/ignite/internal/pagememory/tree/inmemory/ItBplusTreeFakeReuseVolatilePageMemoryTest.java:
##########
@@ -32,7 +32,7 @@ public class ItBplusTreeFakeReuseVolatilePageMemoryTest
extends ItBplusTreeVolat
protected ReuseList createReuseList(
int grpId,
int partId,
- PageMemory pageMem,
+ PartitionPageMemory pageMem,
Review Comment:
1. It can cause a latency spike of up to several hundreds of milliseconds.
In worst cases, on the largest of storage profiles, it can take longer.
2. I have no plans to reduce the time it takes for free-list cache to be
flushed while holding a checkpoint write lock, at least for now. So, I expect
there to be no other overhead besides that (and a write lock usage overhead
itself). The sorting phase (those several hundreds milliseconds) will not
affect the latency anymore.
There was no discussion of changes in Apache Ignite 2 here. That codebase
has a rather different architecture, and other types of optimizations would fit
better into it. Not even mentioning the complexities of a similar refactoring,
it would have been a very complicated task. You can do that yourself if you
like.
4. You can always contribute, right now this particular work is driven by me
alone in my free time.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
