[jira] [Commented] (HADOOP-8499) fix mvn compile -Pnative on CentOS / RHEL / Fedora / SuSE / etc

["Aaron T. Myers (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HADOOP-8499?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13292060#comment-13292060 ] 

Aaron T. Myers commented on HADOOP-8499:
----------------------------------------

bq.  I'm not aware of any Linuxes that create system users with IDs higher than or equal to 500.

Debian and Ubuntu do by default, I believe.
                
> fix mvn compile -Pnative on CentOS / RHEL / Fedora / SuSE / etc
> ---------------------------------------------------------------
>
>                 Key: HADOOP-8499
>                 URL: https://issues.apache.org/jira/browse/HADOOP-8499
>             Project: Hadoop Common
>          Issue Type: Bug
>            Reporter: Colin Patrick McCabe
>            Assignee: Colin Patrick McCabe
>         Attachments: HADOOP-8499.002.patch
>
>
> On Linux platforms where user IDs start at 500 rather than 1000, the build currently is broken.  This includes CentOS, RHEL, Fedora, SuSE, and probably most other Linux platforms.  It does happen to work on Debian and Ubuntu, which explains why Jenkins hasn't caught it yet.
> Other users will see something like this:
> {code}
> [INFO] Requested user cmccabe has id 500, which is below the minimum allowed 1000
> [INFO] FAIL: test-container-executor
> [INFO] ================================================
> [INFO] 1 of 1 test failed
> [INFO] Please report to mapreduce-dev@hadoop.apache.org
> [INFO] ================================================
> [INFO] make[1]: *** [check-TESTS] Error 1
> [INFO] make[1]: Leaving directory `/home/cmccabe/hadoop4/hadoop-mapreduce-project/hadoop-yarn/hadoop-yarn
> -server/hadoop-yarn-server-nodemanager/target/native/container-executor'
> {code}
> And then the build fails.  Since native unit tests are currently unskippable (HADOOP-8480) this makes the project unbuildable.
> The easy solution to this is to relax the constraint for the unit test.  Since the unit test already writes its own configuration file, we just need to change it there.
> In general, I believe that it would make sense to change this to 500 across the board.  I'm not aware of any Linuxes that create system users with IDs higher than or equal to 500.  System user IDs tend to be below 200.
> However, if we do nothing else, we should at least fix the build by relaxing the constraint for unit tests.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

Re: [jira] [Commented] (HADOOP-8499) fix mvn compile -Pnative on CentOS / RHEL / Fedora / SuSE / etc

[Andy Isaacson <ad...@cloudera.com>]

My debian systems start at 100 and allocate upwards.  IIRC there was an old
system of static per-package UID assignments, but as of Debian 6.0 they
appear to be dynamically assigned when you install packages. The highest
UID I have is 116. (Apparently I don't install very many daemon packages.)

Regular user accounts start at 1000 on both Debian and Ubuntu.  As someone
else observed, /etc/login.defs provides the answers:

% egrep '[GU]ID' /etc/login.defs
UID_MIN                  1000
UID_MAX                 60000
#SYS_UID_MIN              100
#SYS_UID_MAX              999
GID_MIN                  1000
GID_MAX                 60000
#SYS_GID_MIN              100
#SYS_GID_MAX              999

-andy