[
https://issues.apache.org/jira/browse/HDFS-124?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12737670#action_12737670
]
Boris Shkolnik commented on HDFS-124:
-------------------------------------
I added currTimeMillis() to the file name.
Since the jira is defined as a performance problems I put here some numbers
before and after the change.
Deleting file with the same name 1000 times. Printing every 100 time. Comparing
with first deletion:
old code:
iteration=100;res =0; start=1249064570781; iterTime = 30 vs. firstTime=26
iteration=200;res =0; start=1249064576620; iterTime = 35 vs. firstTime=26
iteration=300;res =0; start=1249064583456; iterTime = 42 vs. firstTime=26
iteration=400;res =0; start=1249064590637; iterTime = 50 vs. firstTime=26
iteration=500;res =0; start=1249064598285; iterTime = 56 vs. firstTime=26
iteration=600;res =0; start=1249064606626; iterTime = 63 vs. firstTime=26
iteration=700;res =0; start=1249064615643; iterTime = 71 vs. firstTime=26
iteration=800;res =0; start=1249064625248; iterTime = 76 vs. firstTime=26
iteration=900;res =0; start=1249064635654; iterTime = 83 vs. firstTime=26
new code:
iteration=100;res =0; start=1249066273893; iterTime = 22 vs. firstTime=22
iteration=200;res =0; start=1249066278729; iterTime = 22 vs. firstTime=22
iteration=300;res =0; start=1249066283591; iterTime = 22 vs. firstTime=22
iteration=400;res =0; start=1249066288416; iterTime = 24 vs. firstTime=22
iteration=500;res =0; start=1249066293223; iterTime = 27 vs. firstTime=22
iteration=600;res =0; start=1249066298018; iterTime = 30 vs. firstTime=22
iteration=700;res =0; start=1249066302824; iterTime = 25 vs. firstTime=22
iteration=800;res =0; start=1249066307593; iterTime = 23 vs. firstTime=22
iteration=900;res =0; start=1249066312833; iterTime = 23 vs. firstTime=22
> Namenode slowed down when many files with same filename were moved to Trash
> ---------------------------------------------------------------------------
>
> Key: HDFS-124
> URL: https://issues.apache.org/jira/browse/HDFS-124
> Project: Hadoop HDFS
> Issue Type: Bug
> Reporter: Koji Noguchi
> Assignee: Boris Shkolnik
> Priority: Minor
>
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
