[
https://issues.apache.org/jira/browse/HADOOP-18291?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17738895#comment-17738895
]
ASF GitHub Bot commented on HADOOP-18291:
-----------------------------------------
mehakmeet commented on code in PR #5754:
URL: https://github.com/apache/hadoop/pull/5754#discussion_r1247474198
##########
hadoop-tools/hadoop-aws/src/test/java/org/apache/hadoop/fs/s3a/ITestS3APrefetchingInputStream.java:
##########
@@ -301,4 +303,56 @@ public void testStatusProbesAfterClosingStream() throws
Throwable {
}
+ @Test
+ public void testSeeksWithLruEviction() throws Throwable {
+ IOStatistics ioStats;
+ openFS();
+
+ try (FSDataInputStream in = largeFileFS.open(largeFile)) {
+ ioStats = in.getIOStatistics();
+
+ byte[] buffer = new byte[blockSize];
+
+ // Don't read block 0 completely
+ in.read(buffer, 0, blockSize - S_1K * 10);
+
+ // Seek to block 1 and don't read completely
+ in.seek(blockSize);
+ in.read(buffer, 0, 2 * S_1K);
+
+ // Seek to block 2 and don't read completely
+ in.seek(blockSize * 2L);
+ in.read(buffer, 0, 2 * S_1K);
+
+ // Seek to block 3 and don't read completely
+ in.seek(blockSize * 3L);
+ in.read(buffer, 0, 2 * S_1K);
+
+ // Seek to block 4 and don't read completely
+ in.seek(blockSize * 4L);
+ in.read(buffer, 0, 2 * S_1K);
+
+ // Seek to block 5 and don't read completely
+ in.seek(blockSize * 5L);
+ in.read(buffer, 0, 2 * S_1K);
+
+ // backward seek, can't use block 0 as it is evicted
+ in.seek(S_1K * 5);
+ in.read();
+
Review Comment:
Yes, that should be fine as well, I saw this method, so thought we were
already recording that metric `blockRemovedFromFileCache()`
> S3A prefetch - Implement LRU cache for SingleFilePerBlockCache
> --------------------------------------------------------------
>
> Key: HADOOP-18291
> URL: https://issues.apache.org/jira/browse/HADOOP-18291
> Project: Hadoop Common
> Issue Type: Sub-task
> Affects Versions: 3.4.0
> Reporter: Ahmar Suhail
> Assignee: Viraj Jasani
> Priority: Major
> Labels: pull-request-available
>
> Currently there is no limit on the size of disk cache. This means we could
> have a large number of files on files, especially for access patterns that
> are very random and do not always read the block fully.
>
> eg:
> in.seek(5);
> in.read();
> in.seek(blockSize + 10) // block 0 gets saved to disk as it's not fully read
> in.read();
> in.seek(2 * blockSize + 10) // block 1 gets saved to disk
> .. and so on
>
> The in memory cache is bounded, and by default has a limit of 72MB (9
> blocks). When a block is fully read, and a seek is issued it's released
> [here|https://github.com/apache/hadoop/blob/feature-HADOOP-18028-s3a-prefetch/hadoop-tools/hadoop-aws/src/main/java/org/apache/hadoop/fs/s3a/read/S3CachingInputStream.java#L109].
> We can also delete the on disk file for the block here if it exists.
>
> Also maybe add an upper limit on disk space, and delete the file which stores
> data of the block furthest from the current block (similar to the in memory
> cache) when this limit is reached.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
