[
https://issues.apache.org/jira/browse/HTTPCLIENT-2284?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17745926#comment-17745926
]
Oleg Kalnichevski commented on HTTPCLIENT-2284:
-----------------------------------------------
[~lacompas] If I understand it correctly the reported resource leak is about
dangling variant records in the root entry, it is not? If so, in practical
terms it is unlikely to cause any substantial resource or memory issues given
that usually the number of possible response variants is relatively small.
I will look into this issue and try to fix it. If you want to speed up the
process though please do consider contributing a test case reproducing the
issue.
[https://github.com/apache/httpcomponents-client/blob/5.3.x/httpclient5-cache/src/test/java/org/apache/hc/client5/http/impl/cache/TestBasicHttpCache.java]
Oleg
> BasicHttpCacheStorage leaking variant keys in root response's variantMap
> ------------------------------------------------------------------------
>
> Key: HTTPCLIENT-2284
> URL: https://issues.apache.org/jira/browse/HTTPCLIENT-2284
> Project: HttpComponents HttpClient
> Issue Type: Bug
> Components: HttpCache
> Affects Versions: 4.5.13
> Reporter: Pascal Lacombe
> Priority: Major
>
> BasicHttpCacheStorage has a memory leak in the root response's variantMap.
> When a variant cached entry is evicted due to CacheMap being too large, the
> root cache entry does not remove the variant key in its variantMap. This is a
> memory leak that can grow the variantMap indefinitely, or until the root
> entry get's evicted itself.
> Simplified testcase:
> # A request is being sent from a client that contains a header
> "x-my-variant" with a hash of the current timestamp.
> # The server responds 200, with a cacheable response. The response Vary's on
> "x-my-variant"
> # These requests repeat, causing:
> ## The "root" CacheEntry to be kept in CacheMap
> ## The oldest variant CacheEntry to be evicted due to the CacheMap size limit
> ## However the "root" CacheEntry never removes the evicted variant keys from
> the variantMap
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@hc.apache.org
For additional commands, e-mail: dev-h...@hc.apache.org
