Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[SPARK-4864] Add documentation to Netty-based configs #3713

Closed
wants to merge 2 commits into from
+36 −1
Closed

[SPARK-4864] Add documentation to Netty-based configs #3713

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
3b1f84e
[SPARK-4864] Add documentation to Netty-based configs
aarondav Dec 16, 2014
8a8b373
Address Patrick's comments
aarondav Dec 17, 2014
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
53 changes: 53 additions & 0 deletions docs/configuration.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -852,6 +852,59 @@ Apart from these, the following properties are also available, and may be useful
between nodes leading to flooding the network with those.
</td>
</tr>
<tr>
<td><code>spark.shuffle.io.preferDirectBufs</code></td>
<td>true</td>
<td>
(Netty only) Off-heap buffers are used to reduce garbage collection during shuffle and cache
block transfer. For environments where off-heap memory is tightly limited, users may wish to
turn this off to force all allocations from Netty to be on-heap.
</td>
</tr>
<tr>
<td><code>spark.shuffle.io.numConnectionsPerPeer</code></td>
<td>1</td>
<td>
(Netty only) Connections between hosts are reused in order to reduce connection buildup for
large clusters. For small clusters with many hard disks, this may result in insufficient
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

should this say many hard disks and a small number of hosts?

concurrency to saturate all disks, and so users may consider increasing this value.
</td>
</tr>
<tr>
<td><code>spark.shuffle.io.serverThreads</code></td>
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Should this be undocumented? One way I tend to think of it is whether someone could reasonably understand why they should tune this value. It seems a little vague here when someone should adjust this.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Or at least, could you give some indication of exactly why they might increase it (i.e. if you find you can't saturate the network)?

<td>min(num-cores, 8)</td>
<td>
(Netty only) Size of fixed pool of threads to make requests. As each thread gets 16 MB of
off-heap buffer space initially, the default is limited to 8 cores to avoid excessive off-heap
allocation. It has been experimentally determined that this default should be sufficient to
saturate a 10Gb/s network, but in some cases this may have to be adjusted.
</td>
</tr>
<tr>
<td><code>spark.shuffle.io.clientThreads</code></td>
<td>min(num-cores, 8)</td>
<td>
(Netty only) Size of fixed pool of threads to make requests. This has the same properties as
<code>spark.shuffle.io.serverThreads</code>, above.
</td>
</tr>
<tr>
<td><code>spark.shuffle.io.maxRetries</code></td>
<td>3</td>
<td>
(Netty only) Fetches that fail due to IO-related exceptions are automatically retried if this is
set to a non-zero value. This retry logic helps stabilize large shuffles in the face of long GC
pauses or transient network connectivity issues.
</td>
</tr>
<tr>
<td><code>spark.shuffle.io.retryWait</code></td>
<td>5</td>
<td>
(Netty only) Seconds to wait between retries of fetches. The maximum delay caused by retrying
is simply <code>maxRetries * retryWait</code>, by default 15 seconds.
</td>
</tr>
</table>

#### Scheduling
Expand Down
2 changes: 1 addition & 1 deletion network/common/src/main/java/org/apache/spark/network/util/TransportConf.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -40,7 +40,7 @@ public int connectionTimeoutMs() {
return conf.getInt("spark.shuffle.io.connectionTimeout", 120) * 1000;
}

/** Number of concurrent connections between two nodes for fetching data. **/
/** Number of concurrent connections between two nodes for fetching data. */
public int numConnectionsPerPeer() {
return conf.getInt("spark.shuffle.io.numConnectionsPerPeer", 1);
}
Expand Down