[
https://issues.apache.org/jira/browse/ARROW-12846?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17348963#comment-17348963
]
Kouhei Sutou edited comment on ARROW-12846 at 5/21/21, 4:50 AM:
----------------------------------------------------------------
{{dev/release/05-binary-upload.sh}} should be restartable. It should only
upload failed files when we re-run it. If the current
{{dev/release/05-binary-upload.sh}} isn't restartable (is always started from
scratch), it's a bug.
Can I try it? Did you use {{dev/release/05-binary-upload.sh 4.0.1 1}}?
was (Author: kou):
{{dev/release/05-binary-upload.sh}} should be restartable. It should only
upload failed fails when we re-run it. If the current
{{dev/release/05-binary-upload.sh}} isn't restartable (is always started from
scratch), it's a bug.
Can I try it? Did you use {{dev/release/05-binary-upload.sh 4.0.1 1}}?
> [Release] Improve upload of binaries
> ------------------------------------
>
> Key: ARROW-12846
> URL: https://issues.apache.org/jira/browse/ARROW-12846
> Project: Apache Arrow
> Issue Type: Improvement
> Components: Developer Tools
> Reporter: Jorge Leitão
> Assignee: Kouhei Sutou
> Priority: Major
>
> Running dev/release/05-binary-upload.sh takes a long time and is prone to
> network failures, etc. When it fails, it needs to be started from scratch.
> IMO we could alleviate this. An idea here would be to run the script in the
> same order of the configuration variables that it has (e.g.
> `UPLOAD_AMAZON_LINUX_RPM`) and echo the variable when binaries corresponding
> to that section are complete.
> This way, when something fails, as a user I can pass r.g.
> `UPLOAD_AMAZON_LINUX_RPM=0` and skip the parts that were already uploaded.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
