[
https://issues.apache.org/jira/browse/IGNITE-12451?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17203137#comment-17203137
]
Mirza Aliev edited comment on IGNITE-12451 at 9/28/20, 10:21 AM:
-----------------------------------------------------------------
[~alex_pl] Thank you for the detailed explanation! I tend to agree that this
solution is better.
Also, I've discussed the solution with [~agoncharuk] and we agreed that this
solution definitely should be benchmarked as far as atomic cache operations are
lightweight, and even additional adding to HashMap might bring some degradation
on huge batch sizes.
Also, note that _removeAll(Set m)_ operation is also affected
was (Author: maliev):
[~alex_pl] Thank you for the detailed explanation! I tend to agree that this
solution is better.
Also, I've discussed the solution with [~agoncharuk] and we agreed that this
solution definitely should be benchmarked as far as atomic cache operation is
lightweight, and even additional adding to HashMap might bring some degradation
on huge batch sizes.
Also, note that _removeAll(Set m)_ operation is also affected
> Introduce deadlock detection for cache entry reentrant locks
> ------------------------------------------------------------
>
> Key: IGNITE-12451
> URL: https://issues.apache.org/jira/browse/IGNITE-12451
> Project: Ignite
> Issue Type: Improvement
> Affects Versions: 2.7.6
> Reporter: Ivan Rakov
> Assignee: Mirza Aliev
> Priority: Major
> Fix For: 2.10
>
> Time Spent: 20m
> Remaining Estimate: 0h
>
> Aside from IGNITE-12365, we still have possible threat of cache-entry-level
> deadlock in case of careless usage of JCache mass operations (putAll,
> removeAll):
> 1. If two different user threads will perform putAll on the same two keys in
> reverse order (primary node for which is the same), there's a chance that
> sys-stripe threads will be deadlocked.
> 2. Even without direct contract violation from user side, HashMap can be
> passed as argument for putAll. Even if user threads have called mass
> operations with two keys in the same order, HashMap iteration order is not
> strictly defined, which may cause the same deadlock.
> Local deadlock detection should mitigate this issue. We can create a wrapper
> for ReentrantLock with logic that performs cycle detection in wait-for graph
> in case we are waiting for lock acquisition for too long. Exception will be
> thrown from one of the threads in such case, failing user operation, but
> letting the system make progress.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
