infra-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Hervé Boutemy (JIRA) <j...@apache.org>
Subject [jira] [Updated] (INFRA-13213) /x1 disk full on maven-mirror
Date Sun, 18 Jun 2017 11:17:00 GMT

     [ https://issues.apache.org/jira/browse/INFRA-13213?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Hervé Boutemy updated INFRA-13213:
----------------------------------
    Status: Waiting for Infra  (was: Closed)

we had a discussion in Maven PMC: https://lists.apache.org/thread.html/ef71c903099326284f3f06e659855d51eaeac0646ba0a9ddc91f9179@%3Cprivate.maven.apache.org%3E

we got to the conclusion that if we can't afford to have a full copy given its size, having
a copy restricted to metadata would give us some value: ensuring central immutability &
doing partial analysis (although not full analysis, like reindexing full content)

I created a pull request to focus on metadata, ie *.pom, *maven-metadata.xml, *.sha1 and *.asc
https://github.com/apache/infrastructure-puppet/pull/216

You can drop everything else: I don't know if the most efficient will be to re-sync from an
empty filesystem or remove files from existing sync stuck at december 2016

the current estimated size is < 200 GB, with many little files: perhaps the first limit
we'll hit will be inodes count, rather than pure disk size

notice that we expect to create misc analysis from that content, that will require some space
(and more importantly inodes). For example a daily update of https://github.com/hboutemy/mcmm
https://github.com/hboutemy/mcmm-yaml https://github.com/hboutemy/mcmm-logs
then it would be great if we could have some /home space to work on it: once the analysis
is stabilized, we'll probably add it to puppet cron job

> /x1 disk full on maven-mirror
> -----------------------------
>
>                 Key: INFRA-13213
>                 URL: https://issues.apache.org/jira/browse/INFRA-13213
>             Project: Infrastructure
>          Issue Type: Task
>          Components: VM
>            Reporter: Hervé Boutemy
>            Assignee: Chris Lambertus
>
> /x1 is already full: the daily growth is quite high...
> Can you add more space, please?
> you can find daily stats on data synced in
> /home/hboutemy/analysis/central-daily-stats.csv



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

Mime
View raw message