hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Chris Nauroth (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HDFS-8791) block ID-based DN storage layout can be very slow for datanode on ext4
Date Tue, 05 Apr 2016 21:31:25 GMT

    [ https://issues.apache.org/jira/browse/HDFS-8791?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15227167#comment-15227167
] 

Chris Nauroth commented on HDFS-8791:
-------------------------------------

We often see that the layout version has to bump just to support new edit log transactions
for new features.  (Truncate is a great example.)  With HDFS-8432 (introducing the notion
of "minimum compatible layout version"), it's possible to code these features so that while
the upgrade is in progress (unfinalized), attempts to use the new features are rejected. 
Therefore, the new edit log transactions never get written.  Therefore, after a downgrade,
the edit log is still readable by the prior software version.  (No unrecognized/invalid edit
log ops.)  The new features only become usable after the upgrade has been finalized.  At that
point, the admin is announcing intent to stay on "this version" or later.  HDFS-5223 (feature
flags) is not required to achieve this.

HDFS-8432 won't ship until 2.8.0, so unfortunately, it doesn't make our lives easier within
the 2.7 line.  It will help for future major release lines though.

bq. For future 2.7 release to meet this requirement, we could release 2.7.3 that is 2.7.2
plus the DN layout downgrade feature. This is not hard to do.

Maybe I'm missing something, but a DataNode layout downgrade sounds pretty challenging if
it would have to restore older block directory structures by renaming a lot of block and meta
files.

> block ID-based DN storage layout can be very slow for datanode on ext4
> ----------------------------------------------------------------------
>
>                 Key: HDFS-8791
>                 URL: https://issues.apache.org/jira/browse/HDFS-8791
>             Project: Hadoop HDFS
>          Issue Type: Bug
>          Components: datanode
>    Affects Versions: 2.6.0, 2.8.0, 2.7.1
>            Reporter: Nathan Roberts
>            Assignee: Chris Trezzo
>            Priority: Blocker
>             Fix For: 2.7.3
>
>         Attachments: 32x32DatanodeLayoutTesting-v1.pdf, 32x32DatanodeLayoutTesting-v2.pdf,
HDFS-8791-trunk-v1.patch, HDFS-8791-trunk-v2-bin.patch, HDFS-8791-trunk-v2.patch, HDFS-8791-trunk-v2.patch,
HDFS-8791-trunk-v3-bin.patch, hadoop-56-layout-datanode-dir.tgz, test-node-upgrade.txt
>
>
> We are seeing cases where the new directory layout causes the datanode to basically cause
the disks to seek for 10s of minutes. This can be when the datanode is running du, and it
can also be when it is performing a checkDirs(). Both of these operations currently scan all
directories in the block pool and that's very expensive in the new layout.
> The new layout creates 256 subdirs, each with 256 subdirs. Essentially 64K leaf directories
where block files are placed.
> So, what we have on disk is:
> - 256 inodes for the first level directories
> - 256 directory blocks for the first level directories
> - 256*256 inodes for the second level directories
> - 256*256 directory blocks for the second level directories
> - Then the inodes and blocks to store the the HDFS blocks themselves.
> The main problem is the 256*256 directory blocks. 
> inodes and dentries will be cached by linux and one can configure how likely the system
is to prune those entries (vfs_cache_pressure). However, ext4 relies on the buffer cache to
cache the directory blocks and I'm not aware of any way to tell linux to favor buffer cache
pages (even if it did I'm not sure I would want it to in general).
> Also, ext4 tries hard to spread directories evenly across the entire volume, this basically
means the 64K directory blocks are probably randomly spread across the entire disk. A du type
scan will look at directories one at a time, so the ioscheduler can't optimize the corresponding
seeks, meaning the seeks will be random and far. 
> In a system I was using to diagnose this, I had 60K blocks. A DU when things are hot
is less than 1 second. When things are cold, about 20 minutes.
> How do things get cold?
> - A large set of tasks run on the node. This pushes almost all of the buffer cache out,
causing the next DU to hit this situation. We are seeing cases where a large job can cause
a seek storm across the entire cluster.
> Why didn't the previous layout see this?
> - It might have but it wasn't nearly as pronounced. The previous layout would be a few
hundred directory blocks. Even when completely cold, these would only take a few a hundred
seeks which would mean single digit seconds.  
> - With only a few hundred directories, the odds of the directory blocks getting modified
is quite high, this keeps those blocks hot and much less likely to be evicted.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message