[
https://issues.apache.org/jira/browse/HADOOP-3588?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Mahadev konar updated HADOOP-3588:
----------------------------------
Attachment: Hadoop-3588-trunk_1.patch
this patch is for trunk and fixes issues
1, 2, 3, 5, 5.
For 4) I dont think we should disallow it. It is sematically correct to create
a har dir in the input dir and it should be allowed as is allowed in tar or
others. If we are worrying about users to be deleting the directory after they
create the archives then that is something users should be warned about and not
programmed into the code.
For 6) it would need some changes to the Command class that is used by count to
use uris for input. I will open another bug for that.
> Bug report for archives
> -----------------------
>
> Key: HADOOP-3588
> URL: https://issues.apache.org/jira/browse/HADOOP-3588
> Project: Hadoop Core
> Issue Type: Bug
> Components: tools/harchive
> Reporter: Hairong Kuang
> Assignee: Mahadev konar
> Priority: Blocker
> Fix For: 0.18.0
>
> Attachments: Hadoop-3588-trunk_1.patch
>
>
> I tested archives. Here is the bug report:
> Documentation:
> Add examples to section 3 to illustrate the use of listing/cating files in an
> archive.
>
> User interface:
> Case 1:
> $ hadoop archive -archiveName xx.har /conf
> Index: 0, Size: 0
> Case 2: archive does not occur in destination
> $ hadoop archive -archiveName /dd/xx.har /conf /cc
> 08/06/17 22:36:35 INFO mapred.JobClient: Running job: job_200806172141_0005
> 08/06/17 22:36:36 INFO mapred.JobClient: map 0% reduce 0%
> 08/06/17 22:36:39 INFO mapred.JobClient: map 100% reduce 0%
> 08/06/17 22:36:46 INFO mapred.JobClient: Job complete: job_200806172141_0005
> ...
> $ hadoop dfs -ls /dd
> Found 1 items
> drwxr-xr-x - hairong supergroup 0 2008-06-17 22:36 /dd/xx.har
> $ hadoop dfs -ls /cc
> ls: Cannot access /cc: No such file or directory.
> Case 3: archive gets overwritten without any warning
> $ hadoop dfs -ls /dd
> Found 1 items
> drwxr-xr-x - hairong supergroup 0 2008-06-17 22:36 /dd/xx.har
> $ hadoop archive -archiveName xx.har /conf /dd
> 08/06/17 22:43:46 INFO mapred.JobClient: Running job: job_200806172141_0006
> 08/06/17 22:43:47 INFO mapred.JobClient: map 0% reduce 0%
> 08/06/17 22:43:49 INFO mapred.JobClient: map 100% reduce 0%
> 08/06/17 22:43:56 INFO mapred.JobClient: Job complete: job_200806172141_0006
> ...
> $ hadoop dfs -ls /dd
> Found 1 items
> drwxr-xr-x - hairong supergroup 0 2008-06-17 22:43 /dd/xx.har
> Case 4: src & dst are the same - I think we should not allow this
> $ hadoop archive -archiveName xx.har /conf /conf
> 08/06/17 22:52:16 INFO mapred.JobClient: Running job: job_200806172141_0008
> 08/06/17 22:52:17 INFO mapred.JobClient: map 0% reduce 0%
> 08/06/17 22:52:22 INFO mapred.JobClient: map 100% reduce 0%
> 08/06/17 22:52:30 INFO mapred.JobClient: Job complete: job_200806172141_0008
> ...
> $ hadoop dfs -ls /conf
> Found 3 items
> -rw-r--r-- 1 hairong supergroup 3889 2008-06-17 21:43
> /conf/hadoop-site.xml
> -rw-r--r-- 1 hairong supergroup 2844 2008-06-17 21:43
> /conf/log4j.properties
> drwxr-xr-x - hairong supergroup 0 2008-06-17 22:52 /conf/xx.har
> Case 5: dst is a file - more informative message
> $ hadoop dfs -ls /hadoop-site.xml
> Found 1 items
> -rw-r--r-- 1 hairong supergroup 3889 2008-06-17 23:06 /hadoop-site.xml
> $ hadoop archive -archiveName xx.har /conf /hadoop-site.xml
> 08/06/17 23:06:54 INFO mapred.JobClient: Running job: job_200806172141_0009
> Job failed!
> Functionality:
> Case 5: error message is better to be "op" not allowed.
> $ hadoop dfs -ls har:///test/xx.har/conf
> Found 2 items
> -rw-r--r-- 10 hairong supergroup 3889 2008-06-17 23:24
> /test/xx.har/conf/hadoop-site.xml
> -rw-r--r-- 10 hairong supergroup 2844 2008-06-17 23:24
> /test/xx.har/conf/log4j.properties
> $ hadoop dfs -ls har:///test/xx.har/conf/hadoop-site.xml
> Found 1 items
> -rw-r--r-- 10 hairong supergroup 3889 2008-06-17 23:24
> /test/xx.har/conf/hadoop-site.xml
> $ hadoop dfs -rm har:///test/xx.har/conf/hadoop-site.xml
> rm: Har: delete not implemented
> $ hadoop dfs -mv har:///test/xx.har/conf/hadoop-site.xml
> har:///test/xx.har/conf/hadoop-default.xml
> mv: Failed to rename har:/test/xx.har/conf/hadoop-site.xml to
> har:/test/xx.har/conf/hadoop-default.xml
> Case 6: count is not implemented
> $ hadoop dfs -count har:///test/xx.har/conf
> Can not find listing for har:///test/xx.har/conf
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
