You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ratis.apache.org by "Attila Doroszlai (Jira)" <ji...@apache.org> on 2026/10/02 06:28:00 UTC

[jira] [Updated] (RATIS-2725) Tune maven-resolver HTTP retry settings for CI

     [ https://issues.apache.org/jira/browse/RATIS-2725?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Attila Doroszlai updated RATIS-2725:
------------------------------------
    Fix Version/s: 3.4.0

> Tune maven-resolver HTTP retry settings for CI
> ----------------------------------------------
>
>                 Key: RATIS-2725
>                 URL: https://issues.apache.org/jira/browse/RATIS-2725
>             Project: Ratis
>          Issue Type: Task
>          Components: CI
>            Reporter: Dian-Xuan Yang
>            Assignee: Dian-Xuan Yang
>            Priority: Major
>             Fix For: 3.4.0
>
>          Time Spent: 50m
>  Remaining Estimate: 0h
>
> RATIS-2698 mirrored the retry settings from HDDS-16359. HDDS-16648 later found those values too aggressive. With {{retryHandler.count=10}}, a single download that fails with 429 is retried for about 275s (11 requests). Each retry adds pressure on Maven Central, which is already asking clients to slow down. Also, {{-Dhttp.keepAlive=false}} has no effect: the resolver's HttpClient does not read it.
> Apply the same settings as HDDS-16648 in both ratis and ratis-thirdparty:
> * {{retryHandler.count}} from 10 to 5
> * add {{retryHandler.intervalMax=30000}}, so the build fails fast when {{Retry-After}} exceeds 30s
> * add {{connectionMaxTtl=25}}
> * remove {{-Dhttp.keepAlive=false}}
> Measured with Maven 3.9.16 against a mirror that always returns 429 (see [apache/ozone#11375|https://github.com/apache/ozone/pull/11375]):
> ||Settings||Retry-After||Requests||Elapsed||
> |Current|none|11|277.4s|
> |New|none|6|77.0s|
> |Current|60|11|602.4s|
> |New|60|1|2.3s|



--
This message was sent by Atlassian Jira
(v8.20.10#820010)