[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Steve Loughran updated HADOOP-10400: Assignee: Jordan Mendelson Incorporate new S3A FileSystem implementation - Key: HADOOP-10400 URL: https://issues.apache.org/jira/browse/HADOOP-10400 Project: Hadoop Common Issue Type: New Feature Components: fs, fs/s3 Affects Versions: 2.4.0 Reporter: Jordan Mendelson Assignee: Jordan Mendelson Fix For: 2.6.0 Attachments: HADOOP-10400-1.patch, HADOOP-10400-2.patch, HADOOP-10400-3.patch, HADOOP-10400-4.patch, HADOOP-10400-5.patch, HADOOP-10400-6.patch, HADOOP-10400-7.patch, HADOOP-10400-8-branch-2.patch, HADOOP-10400-8.patch, HADOOP-10400-branch-2.patch The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Tunable parameters:* fs.s3a.access.key - Your AWS access key ID (omit for role authentication) fs.s3a.secret.key - Your AWS secret key (omit for role authentication) fs.s3a.connection.maximum - Controls how many parallel connections HttpClient spawns (default: 15) fs.s3a.connection.ssl.enabled - Enables or disables SSL connections to S3 (default: true) fs.s3a.attempts.maximum - How many times we should retry commands on transient errors (default: 10) fs.s3a.connection.timeout - Socket connect timeout (default: 5000) fs.s3a.paging.maximum - How many keys to request from S3 when doing directory listings at a time (default: 5000) fs.s3a.multipart.size - How big (in bytes) to split a upload or copy operation up into (default: 104857600) fs.s3a.multipart.threshold - Until a file is this large (in bytes), use non-parallel upload (default: 2147483647) fs.s3a.acl.default - Set a canned ACL on newly created/copied objects (private | public-read | public-read-write | authenticated-read | log-delivery-write | bucket-owner-read | bucket-owner-full-control) fs.s3a.multipart.purge - True if you want to purge existing multipart uploads that may not have been completed/aborted correctly (default: false) fs.s3a.multipart.purge.age - Minimum age in seconds of multipart uploads to purge (default: 86400) fs.s3a.buffer.dir - Comma separated list of directories that will be used to buffer file writes out of (default: uses ${hadoop.tmp.dir}/s3a ) *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length and MD5 to be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is still being made for rename operations, so your job may time out unless you increase the task timeout. This driver will fully ignore _$folder$ files. This was necessary so that it could interoperate with repositories that have had the s3native driver used on them, but means that it won't recognize empty directories that s3native has been used on. Statistics for the filesystem may
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Chris Bannister updated HADOOP-10400: - Assignee: (was: Chris Bannister) Incorporate new S3A FileSystem implementation - Key: HADOOP-10400 URL: https://issues.apache.org/jira/browse/HADOOP-10400 Project: Hadoop Common Issue Type: New Feature Components: fs, fs/s3 Affects Versions: 2.4.0 Reporter: Jordan Mendelson Fix For: 2.6.0 Attachments: HADOOP-10400-1.patch, HADOOP-10400-2.patch, HADOOP-10400-3.patch, HADOOP-10400-4.patch, HADOOP-10400-5.patch, HADOOP-10400-6.patch, HADOOP-10400-7.patch, HADOOP-10400-8-branch-2.patch, HADOOP-10400-8.patch, HADOOP-10400-branch-2.patch The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Tunable parameters:* fs.s3a.access.key - Your AWS access key ID (omit for role authentication) fs.s3a.secret.key - Your AWS secret key (omit for role authentication) fs.s3a.connection.maximum - Controls how many parallel connections HttpClient spawns (default: 15) fs.s3a.connection.ssl.enabled - Enables or disables SSL connections to S3 (default: true) fs.s3a.attempts.maximum - How many times we should retry commands on transient errors (default: 10) fs.s3a.connection.timeout - Socket connect timeout (default: 5000) fs.s3a.paging.maximum - How many keys to request from S3 when doing directory listings at a time (default: 5000) fs.s3a.multipart.size - How big (in bytes) to split a upload or copy operation up into (default: 104857600) fs.s3a.multipart.threshold - Until a file is this large (in bytes), use non-parallel upload (default: 2147483647) fs.s3a.acl.default - Set a canned ACL on newly created/copied objects (private | public-read | public-read-write | authenticated-read | log-delivery-write | bucket-owner-read | bucket-owner-full-control) fs.s3a.multipart.purge - True if you want to purge existing multipart uploads that may not have been completed/aborted correctly (default: false) fs.s3a.multipart.purge.age - Minimum age in seconds of multipart uploads to purge (default: 86400) fs.s3a.buffer.dir - Comma separated list of directories that will be used to buffer file writes out of (default: uses ${hadoop.tmp.dir}/s3a ) *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length and MD5 to be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is still being made for rename operations, so your job may time out unless you increase the task timeout. This driver will fully ignore _$folder$ files. This was necessary so that it could interoperate with repositories that have had the s3native driver used on them, but means that it won't recognize empty directories that s3native has been used on. Statistics for the filesystem may be calculated differently
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Aaron T. Myers updated HADOOP-10400: Issue Type: New Feature (was: Improvement) Incorporate new S3A FileSystem implementation - Key: HADOOP-10400 URL: https://issues.apache.org/jira/browse/HADOOP-10400 Project: Hadoop Common Issue Type: New Feature Components: fs, fs/s3 Affects Versions: 2.4.0 Reporter: Jordan Mendelson Assignee: Jordan Mendelson Attachments: HADOOP-10400-1.patch, HADOOP-10400-2.patch, HADOOP-10400-3.patch, HADOOP-10400-4.patch, HADOOP-10400-5.patch, HADOOP-10400-6.patch, HADOOP-10400-7.patch, HADOOP-10400-8-branch-2.patch, HADOOP-10400-8.patch, HADOOP-10400-branch-2.patch The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Tunable parameters:* fs.s3a.access.key - Your AWS access key ID (omit for role authentication) fs.s3a.secret.key - Your AWS secret key (omit for role authentication) fs.s3a.connection.maximum - Controls how many parallel connections HttpClient spawns (default: 15) fs.s3a.connection.ssl.enabled - Enables or disables SSL connections to S3 (default: true) fs.s3a.attempts.maximum - How many times we should retry commands on transient errors (default: 10) fs.s3a.connection.timeout - Socket connect timeout (default: 5000) fs.s3a.paging.maximum - How many keys to request from S3 when doing directory listings at a time (default: 5000) fs.s3a.multipart.size - How big (in bytes) to split a upload or copy operation up into (default: 104857600) fs.s3a.multipart.threshold - Until a file is this large (in bytes), use non-parallel upload (default: 2147483647) fs.s3a.acl.default - Set a canned ACL on newly created/copied objects (private | public-read | public-read-write | authenticated-read | log-delivery-write | bucket-owner-read | bucket-owner-full-control) fs.s3a.multipart.purge - True if you want to purge existing multipart uploads that may not have been completed/aborted correctly (default: false) fs.s3a.multipart.purge.age - Minimum age in seconds of multipart uploads to purge (default: 86400) fs.s3a.buffer.dir - Comma separated list of directories that will be used to buffer file writes out of (default: uses ${hadoop.tmp.dir}/s3a ) *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length and MD5 to be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is still being made for rename operations, so your job may time out unless you increase the task timeout. This driver will fully ignore _$folder$ files. This was necessary so that it could interoperate with repositories that have had the s3native driver used on them, but means that it won't recognize empty directories that s3native has been used on. Statistics for the filesystem may be calculated
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Aaron T. Myers updated HADOOP-10400: Resolution: Fixed Fix Version/s: 2.6.0 Target Version/s: 2.6.0 (was: 3.0.0) Hadoop Flags: Reviewed Status: Resolved (was: Patch Available) I've just committed this to trunk and branch-2. Thanks a lot, Jordan, for the initial implementation and thanks to Dave for taking the patch over the finish line. Thanks also to Steve and others for all of the reviews they've provided. Incorporate new S3A FileSystem implementation - Key: HADOOP-10400 URL: https://issues.apache.org/jira/browse/HADOOP-10400 Project: Hadoop Common Issue Type: New Feature Components: fs, fs/s3 Affects Versions: 2.4.0 Reporter: Jordan Mendelson Assignee: Jordan Mendelson Fix For: 2.6.0 Attachments: HADOOP-10400-1.patch, HADOOP-10400-2.patch, HADOOP-10400-3.patch, HADOOP-10400-4.patch, HADOOP-10400-5.patch, HADOOP-10400-6.patch, HADOOP-10400-7.patch, HADOOP-10400-8-branch-2.patch, HADOOP-10400-8.patch, HADOOP-10400-branch-2.patch The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Tunable parameters:* fs.s3a.access.key - Your AWS access key ID (omit for role authentication) fs.s3a.secret.key - Your AWS secret key (omit for role authentication) fs.s3a.connection.maximum - Controls how many parallel connections HttpClient spawns (default: 15) fs.s3a.connection.ssl.enabled - Enables or disables SSL connections to S3 (default: true) fs.s3a.attempts.maximum - How many times we should retry commands on transient errors (default: 10) fs.s3a.connection.timeout - Socket connect timeout (default: 5000) fs.s3a.paging.maximum - How many keys to request from S3 when doing directory listings at a time (default: 5000) fs.s3a.multipart.size - How big (in bytes) to split a upload or copy operation up into (default: 104857600) fs.s3a.multipart.threshold - Until a file is this large (in bytes), use non-parallel upload (default: 2147483647) fs.s3a.acl.default - Set a canned ACL on newly created/copied objects (private | public-read | public-read-write | authenticated-read | log-delivery-write | bucket-owner-read | bucket-owner-full-control) fs.s3a.multipart.purge - True if you want to purge existing multipart uploads that may not have been completed/aborted correctly (default: false) fs.s3a.multipart.purge.age - Minimum age in seconds of multipart uploads to purge (default: 86400) fs.s3a.buffer.dir - Comma separated list of directories that will be used to buffer file writes out of (default: uses ${hadoop.tmp.dir}/s3a ) *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length and MD5 to be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] David S. Wang updated HADOOP-10400: --- Attachment: HADOOP-10400-8-branch-2.patch Here's a branch-2 backport for the latest patch HADOOP-10400-8.patch. Incorporate new S3A FileSystem implementation - Key: HADOOP-10400 URL: https://issues.apache.org/jira/browse/HADOOP-10400 Project: Hadoop Common Issue Type: Improvement Components: fs, fs/s3 Affects Versions: 2.4.0 Reporter: Jordan Mendelson Assignee: Jordan Mendelson Attachments: HADOOP-10400-1.patch, HADOOP-10400-2.patch, HADOOP-10400-3.patch, HADOOP-10400-4.patch, HADOOP-10400-5.patch, HADOOP-10400-6.patch, HADOOP-10400-7.patch, HADOOP-10400-8-branch-2.patch, HADOOP-10400-8.patch, HADOOP-10400-branch-2.patch The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Tunable parameters:* fs.s3a.access.key - Your AWS access key ID (omit for role authentication) fs.s3a.secret.key - Your AWS secret key (omit for role authentication) fs.s3a.connection.maximum - Controls how many parallel connections HttpClient spawns (default: 15) fs.s3a.connection.ssl.enabled - Enables or disables SSL connections to S3 (default: true) fs.s3a.attempts.maximum - How many times we should retry commands on transient errors (default: 10) fs.s3a.connection.timeout - Socket connect timeout (default: 5000) fs.s3a.paging.maximum - How many keys to request from S3 when doing directory listings at a time (default: 5000) fs.s3a.multipart.size - How big (in bytes) to split a upload or copy operation up into (default: 104857600) fs.s3a.multipart.threshold - Until a file is this large (in bytes), use non-parallel upload (default: 2147483647) fs.s3a.acl.default - Set a canned ACL on newly created/copied objects (private | public-read | public-read-write | authenticated-read | log-delivery-write | bucket-owner-read | bucket-owner-full-control) fs.s3a.multipart.purge - True if you want to purge existing multipart uploads that may not have been completed/aborted correctly (default: false) fs.s3a.multipart.purge.age - Minimum age in seconds of multipart uploads to purge (default: 86400) fs.s3a.buffer.dir - Comma separated list of directories that will be used to buffer file writes out of (default: uses ${hadoop.tmp.dir}/s3a ) *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length and MD5 to be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is still being made for rename operations, so your job may time out unless you increase the task timeout. This driver will fully ignore _$folder$ files. This was necessary so that it could interoperate with repositories that have had the s3native driver used on them, but means that it won't recognize empty directories that s3native
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] David S. Wang updated HADOOP-10400: --- Status: Open (was: Patch Available) Incorporate new S3A FileSystem implementation - Key: HADOOP-10400 URL: https://issues.apache.org/jira/browse/HADOOP-10400 Project: Hadoop Common Issue Type: Improvement Components: fs, fs/s3 Affects Versions: 2.4.0 Reporter: Jordan Mendelson Assignee: Jordan Mendelson Attachments: HADOOP-10400-1.patch, HADOOP-10400-2.patch, HADOOP-10400-3.patch, HADOOP-10400-4.patch, HADOOP-10400-5.patch, HADOOP-10400-6.patch, HADOOP-10400-7.patch, HADOOP-10400-branch-2.patch The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Tunable parameters:* fs.s3a.access.key - Your AWS access key ID (omit for role authentication) fs.s3a.secret.key - Your AWS secret key (omit for role authentication) fs.s3a.connection.maximum - Controls how many parallel connections HttpClient spawns (default: 15) fs.s3a.connection.ssl.enabled - Enables or disables SSL connections to S3 (default: true) fs.s3a.attempts.maximum - How many times we should retry commands on transient errors (default: 10) fs.s3a.connection.timeout - Socket connect timeout (default: 5000) fs.s3a.paging.maximum - How many keys to request from S3 when doing directory listings at a time (default: 5000) fs.s3a.multipart.size - How big (in bytes) to split a upload or copy operation up into (default: 104857600) fs.s3a.multipart.threshold - Until a file is this large (in bytes), use non-parallel upload (default: 2147483647) fs.s3a.acl.default - Set a canned ACL on newly created/copied objects (private | public-read | public-read-write | authenticated-read | log-delivery-write | bucket-owner-read | bucket-owner-full-control) fs.s3a.multipart.purge - True if you want to purge existing multipart uploads that may not have been completed/aborted correctly (default: false) fs.s3a.multipart.purge.age - Minimum age in seconds of multipart uploads to purge (default: 86400) fs.s3a.buffer.dir - Comma separated list of directories that will be used to buffer file writes out of (default: uses ${hadoop.tmp.dir}/s3a ) *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length and MD5 to be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is still being made for rename operations, so your job may time out unless you increase the task timeout. This driver will fully ignore _$folder$ files. This was necessary so that it could interoperate with repositories that have had the s3native driver used on them, but means that it won't recognize empty directories that s3native has been used on. Statistics for the filesystem may be calculated differently than the s3native filesystem. When uploading a
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] David S. Wang updated HADOOP-10400: --- Attachment: HADOOP-10400-8.patch Thanks Steve for your comments. I've attached a trunk patch to address your non-longer-term concerns. I believe all of them are addressed. I did the same testing as I mentioned as in previous comments. In addition, all of the newly-added s3a FS contract tests pass. Incorporate new S3A FileSystem implementation - Key: HADOOP-10400 URL: https://issues.apache.org/jira/browse/HADOOP-10400 Project: Hadoop Common Issue Type: Improvement Components: fs, fs/s3 Affects Versions: 2.4.0 Reporter: Jordan Mendelson Assignee: Jordan Mendelson Attachments: HADOOP-10400-1.patch, HADOOP-10400-2.patch, HADOOP-10400-3.patch, HADOOP-10400-4.patch, HADOOP-10400-5.patch, HADOOP-10400-6.patch, HADOOP-10400-7.patch, HADOOP-10400-8.patch, HADOOP-10400-branch-2.patch The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Tunable parameters:* fs.s3a.access.key - Your AWS access key ID (omit for role authentication) fs.s3a.secret.key - Your AWS secret key (omit for role authentication) fs.s3a.connection.maximum - Controls how many parallel connections HttpClient spawns (default: 15) fs.s3a.connection.ssl.enabled - Enables or disables SSL connections to S3 (default: true) fs.s3a.attempts.maximum - How many times we should retry commands on transient errors (default: 10) fs.s3a.connection.timeout - Socket connect timeout (default: 5000) fs.s3a.paging.maximum - How many keys to request from S3 when doing directory listings at a time (default: 5000) fs.s3a.multipart.size - How big (in bytes) to split a upload or copy operation up into (default: 104857600) fs.s3a.multipart.threshold - Until a file is this large (in bytes), use non-parallel upload (default: 2147483647) fs.s3a.acl.default - Set a canned ACL on newly created/copied objects (private | public-read | public-read-write | authenticated-read | log-delivery-write | bucket-owner-read | bucket-owner-full-control) fs.s3a.multipart.purge - True if you want to purge existing multipart uploads that may not have been completed/aborted correctly (default: false) fs.s3a.multipart.purge.age - Minimum age in seconds of multipart uploads to purge (default: 86400) fs.s3a.buffer.dir - Comma separated list of directories that will be used to buffer file writes out of (default: uses ${hadoop.tmp.dir}/s3a ) *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length and MD5 to be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is still being made for rename operations, so your job may time out unless you increase the task timeout. This driver will fully ignore _$folder$ files. This was necessary so
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] David S. Wang updated HADOOP-10400: --- Status: Patch Available (was: Open) Incorporate new S3A FileSystem implementation - Key: HADOOP-10400 URL: https://issues.apache.org/jira/browse/HADOOP-10400 Project: Hadoop Common Issue Type: Improvement Components: fs, fs/s3 Affects Versions: 2.4.0 Reporter: Jordan Mendelson Assignee: Jordan Mendelson Attachments: HADOOP-10400-1.patch, HADOOP-10400-2.patch, HADOOP-10400-3.patch, HADOOP-10400-4.patch, HADOOP-10400-5.patch, HADOOP-10400-6.patch, HADOOP-10400-7.patch, HADOOP-10400-8.patch, HADOOP-10400-branch-2.patch The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Tunable parameters:* fs.s3a.access.key - Your AWS access key ID (omit for role authentication) fs.s3a.secret.key - Your AWS secret key (omit for role authentication) fs.s3a.connection.maximum - Controls how many parallel connections HttpClient spawns (default: 15) fs.s3a.connection.ssl.enabled - Enables or disables SSL connections to S3 (default: true) fs.s3a.attempts.maximum - How many times we should retry commands on transient errors (default: 10) fs.s3a.connection.timeout - Socket connect timeout (default: 5000) fs.s3a.paging.maximum - How many keys to request from S3 when doing directory listings at a time (default: 5000) fs.s3a.multipart.size - How big (in bytes) to split a upload or copy operation up into (default: 104857600) fs.s3a.multipart.threshold - Until a file is this large (in bytes), use non-parallel upload (default: 2147483647) fs.s3a.acl.default - Set a canned ACL on newly created/copied objects (private | public-read | public-read-write | authenticated-read | log-delivery-write | bucket-owner-read | bucket-owner-full-control) fs.s3a.multipart.purge - True if you want to purge existing multipart uploads that may not have been completed/aborted correctly (default: false) fs.s3a.multipart.purge.age - Minimum age in seconds of multipart uploads to purge (default: 86400) fs.s3a.buffer.dir - Comma separated list of directories that will be used to buffer file writes out of (default: uses ${hadoop.tmp.dir}/s3a ) *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length and MD5 to be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is still being made for rename operations, so your job may time out unless you increase the task timeout. This driver will fully ignore _$folder$ files. This was necessary so that it could interoperate with repositories that have had the s3native driver used on them, but means that it won't recognize empty directories that s3native has been used on. Statistics for the filesystem may be calculated differently than the s3native
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] David S. Wang updated HADOOP-10400: --- Status: Open (was: Patch Available) Incorporate new S3A FileSystem implementation - Key: HADOOP-10400 URL: https://issues.apache.org/jira/browse/HADOOP-10400 Project: Hadoop Common Issue Type: Improvement Components: fs, fs/s3 Affects Versions: 2.4.0 Reporter: Jordan Mendelson Assignee: Jordan Mendelson Attachments: HADOOP-10400-1.patch, HADOOP-10400-2.patch, HADOOP-10400-3.patch, HADOOP-10400-4.patch, HADOOP-10400-5.patch, HADOOP-10400-6.patch The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Tunable parameters:* fs.s3a.access.key - Your AWS access key ID (omit for role authentication) fs.s3a.secret.key - Your AWS secret key (omit for role authentication) fs.s3a.connection.maximum - Controls how many parallel connections HttpClient spawns (default: 15) fs.s3a.connection.ssl.enabled - Enables or disables SSL connections to S3 (default: true) fs.s3a.attempts.maximum - How many times we should retry commands on transient errors (default: 10) fs.s3a.connection.timeout - Socket connect timeout (default: 5000) fs.s3a.paging.maximum - How many keys to request from S3 when doing directory listings at a time (default: 5000) fs.s3a.multipart.size - How big (in bytes) to split a upload or copy operation up into (default: 104857600) fs.s3a.multipart.threshold - Until a file is this large (in bytes), use non-parallel upload (default: 2147483647) fs.s3a.acl.default - Set a canned ACL on newly created/copied objects (private | public-read | public-read-write | authenticated-read | log-delivery-write | bucket-owner-read | bucket-owner-full-control) fs.s3a.multipart.purge - True if you want to purge existing multipart uploads that may not have been completed/aborted correctly (default: false) fs.s3a.multipart.purge.age - Minimum age in seconds of multipart uploads to purge (default: 86400) fs.s3a.buffer.dir - Comma separated list of directories that will be used to buffer file writes out of (default: uses ${hadoop.tmp.dir}/s3a ) *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length and MD5 to be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is still being made for rename operations, so your job may time out unless you increase the task timeout. This driver will fully ignore _$folder$ files. This was necessary so that it could interoperate with repositories that have had the s3native driver used on them, but means that it won't recognize empty directories that s3native has been used on. Statistics for the filesystem may be calculated differently than the s3native filesystem. When uploading a file, we do not count writing the temporary file on
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] David S. Wang updated HADOOP-10400: --- Attachment: HADOOP-10400-7.patch HADOOP-10400-7.patch does the following: * Rebases HADOOP-10400-6.patch onto current tip-of-trunk, which includes the changes to HADOOP-11074 to move the s3 connector bits over to hadoop-aws. * Incorporates HADOOP-10675, HADOOP-10676, and HADOOP-10677, which were fixes on top of previous candidate HADOOP-10400 patches. * Corrects jackson 2 dependencies used by the hadoop-aws and hadoop-azure modules. With regards to testing, I ran mvn clean install -Pnative -DskipTests from top-level, and mvn test in both hadoop-aws and hadoop-azure directories. Incorporate new S3A FileSystem implementation - Key: HADOOP-10400 URL: https://issues.apache.org/jira/browse/HADOOP-10400 Project: Hadoop Common Issue Type: Improvement Components: fs, fs/s3 Affects Versions: 2.4.0 Reporter: Jordan Mendelson Assignee: Jordan Mendelson Attachments: HADOOP-10400-1.patch, HADOOP-10400-2.patch, HADOOP-10400-3.patch, HADOOP-10400-4.patch, HADOOP-10400-5.patch, HADOOP-10400-6.patch, HADOOP-10400-7.patch The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Tunable parameters:* fs.s3a.access.key - Your AWS access key ID (omit for role authentication) fs.s3a.secret.key - Your AWS secret key (omit for role authentication) fs.s3a.connection.maximum - Controls how many parallel connections HttpClient spawns (default: 15) fs.s3a.connection.ssl.enabled - Enables or disables SSL connections to S3 (default: true) fs.s3a.attempts.maximum - How many times we should retry commands on transient errors (default: 10) fs.s3a.connection.timeout - Socket connect timeout (default: 5000) fs.s3a.paging.maximum - How many keys to request from S3 when doing directory listings at a time (default: 5000) fs.s3a.multipart.size - How big (in bytes) to split a upload or copy operation up into (default: 104857600) fs.s3a.multipart.threshold - Until a file is this large (in bytes), use non-parallel upload (default: 2147483647) fs.s3a.acl.default - Set a canned ACL on newly created/copied objects (private | public-read | public-read-write | authenticated-read | log-delivery-write | bucket-owner-read | bucket-owner-full-control) fs.s3a.multipart.purge - True if you want to purge existing multipart uploads that may not have been completed/aborted correctly (default: false) fs.s3a.multipart.purge.age - Minimum age in seconds of multipart uploads to purge (default: 86400) fs.s3a.buffer.dir - Comma separated list of directories that will be used to buffer file writes out of (default: uses ${hadoop.tmp.dir}/s3a ) *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length and MD5 to be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while.
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] David S. Wang updated HADOOP-10400: --- Target Version/s: 3.0.0 Status: Patch Available (was: Open) Incorporate new S3A FileSystem implementation - Key: HADOOP-10400 URL: https://issues.apache.org/jira/browse/HADOOP-10400 Project: Hadoop Common Issue Type: Improvement Components: fs, fs/s3 Affects Versions: 2.4.0 Reporter: Jordan Mendelson Assignee: Jordan Mendelson Attachments: HADOOP-10400-1.patch, HADOOP-10400-2.patch, HADOOP-10400-3.patch, HADOOP-10400-4.patch, HADOOP-10400-5.patch, HADOOP-10400-6.patch, HADOOP-10400-7.patch The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Tunable parameters:* fs.s3a.access.key - Your AWS access key ID (omit for role authentication) fs.s3a.secret.key - Your AWS secret key (omit for role authentication) fs.s3a.connection.maximum - Controls how many parallel connections HttpClient spawns (default: 15) fs.s3a.connection.ssl.enabled - Enables or disables SSL connections to S3 (default: true) fs.s3a.attempts.maximum - How many times we should retry commands on transient errors (default: 10) fs.s3a.connection.timeout - Socket connect timeout (default: 5000) fs.s3a.paging.maximum - How many keys to request from S3 when doing directory listings at a time (default: 5000) fs.s3a.multipart.size - How big (in bytes) to split a upload or copy operation up into (default: 104857600) fs.s3a.multipart.threshold - Until a file is this large (in bytes), use non-parallel upload (default: 2147483647) fs.s3a.acl.default - Set a canned ACL on newly created/copied objects (private | public-read | public-read-write | authenticated-read | log-delivery-write | bucket-owner-read | bucket-owner-full-control) fs.s3a.multipart.purge - True if you want to purge existing multipart uploads that may not have been completed/aborted correctly (default: false) fs.s3a.multipart.purge.age - Minimum age in seconds of multipart uploads to purge (default: 86400) fs.s3a.buffer.dir - Comma separated list of directories that will be used to buffer file writes out of (default: uses ${hadoop.tmp.dir}/s3a ) *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length and MD5 to be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is still being made for rename operations, so your job may time out unless you increase the task timeout. This driver will fully ignore _$folder$ files. This was necessary so that it could interoperate with repositories that have had the s3native driver used on them, but means that it won't recognize empty directories that s3native has been used on. Statistics for the filesystem may be calculated differently than the s3native filesystem. When
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] David S. Wang updated HADOOP-10400: --- Attachment: HADOOP-10400-branch-2.patch This is the branch-2 backport for HADOOP-10400. It relies on HADOOP-11074 to be applied first. It applied almost entirely cleanly, except for a reference to hadoop-azure in a POM file, which is not in branch-2. Incorporate new S3A FileSystem implementation - Key: HADOOP-10400 URL: https://issues.apache.org/jira/browse/HADOOP-10400 Project: Hadoop Common Issue Type: Improvement Components: fs, fs/s3 Affects Versions: 2.4.0 Reporter: Jordan Mendelson Assignee: Jordan Mendelson Attachments: HADOOP-10400-1.patch, HADOOP-10400-2.patch, HADOOP-10400-3.patch, HADOOP-10400-4.patch, HADOOP-10400-5.patch, HADOOP-10400-6.patch, HADOOP-10400-7.patch, HADOOP-10400-branch-2.patch The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Tunable parameters:* fs.s3a.access.key - Your AWS access key ID (omit for role authentication) fs.s3a.secret.key - Your AWS secret key (omit for role authentication) fs.s3a.connection.maximum - Controls how many parallel connections HttpClient spawns (default: 15) fs.s3a.connection.ssl.enabled - Enables or disables SSL connections to S3 (default: true) fs.s3a.attempts.maximum - How many times we should retry commands on transient errors (default: 10) fs.s3a.connection.timeout - Socket connect timeout (default: 5000) fs.s3a.paging.maximum - How many keys to request from S3 when doing directory listings at a time (default: 5000) fs.s3a.multipart.size - How big (in bytes) to split a upload or copy operation up into (default: 104857600) fs.s3a.multipart.threshold - Until a file is this large (in bytes), use non-parallel upload (default: 2147483647) fs.s3a.acl.default - Set a canned ACL on newly created/copied objects (private | public-read | public-read-write | authenticated-read | log-delivery-write | bucket-owner-read | bucket-owner-full-control) fs.s3a.multipart.purge - True if you want to purge existing multipart uploads that may not have been completed/aborted correctly (default: false) fs.s3a.multipart.purge.age - Minimum age in seconds of multipart uploads to purge (default: 86400) fs.s3a.buffer.dir - Comma separated list of directories that will be used to buffer file writes out of (default: uses ${hadoop.tmp.dir}/s3a ) *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length and MD5 to be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is still being made for rename operations, so your job may time out unless you increase the task timeout. This driver will fully ignore _$folder$ files. This was necessary so that it could interoperate with repositories that have had the s3native
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Matteo Bertozzi updated HADOOP-10400: - Attachment: HADOOP-10400-6.patch attached v6, which is the same as v5 but fixes the fs.open().close() case. The wrappedObject is initialized only inside the read, so calling close() before a read will throw an NPE. testInputStreamClosedTwice() should reproduce the problem, since is doing fs.open().close() {code} @@ -175,7 +175,9 @@ public class S3AInputStream extends FSInputStream { } super.close(); closed = true; -wrappedObject.close(); +if (wrappedObject != null) { + wrappedObject.close(); +} } {code} Incorporate new S3A FileSystem implementation - Key: HADOOP-10400 URL: https://issues.apache.org/jira/browse/HADOOP-10400 Project: Hadoop Common Issue Type: Improvement Components: fs, fs/s3 Affects Versions: 2.4.0 Reporter: Jordan Mendelson Assignee: Jordan Mendelson Attachments: HADOOP-10400-1.patch, HADOOP-10400-2.patch, HADOOP-10400-3.patch, HADOOP-10400-4.patch, HADOOP-10400-5.patch, HADOOP-10400-6.patch The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Tunable parameters:* fs.s3a.access.key - Your AWS access key ID (omit for role authentication) fs.s3a.secret.key - Your AWS secret key (omit for role authentication) fs.s3a.connection.maximum - Controls how many parallel connections HttpClient spawns (default: 15) fs.s3a.connection.ssl.enabled - Enables or disables SSL connections to S3 (default: true) fs.s3a.attempts.maximum - How many times we should retry commands on transient errors (default: 10) fs.s3a.connection.timeout - Socket connect timeout (default: 5000) fs.s3a.paging.maximum - How many keys to request from S3 when doing directory listings at a time (default: 5000) fs.s3a.multipart.size - How big (in bytes) to split a upload or copy operation up into (default: 104857600) fs.s3a.multipart.threshold - Until a file is this large (in bytes), use non-parallel upload (default: 2147483647) fs.s3a.acl.default - Set a canned ACL on newly created/copied objects (private | public-read | public-read-write | authenticated-read | log-delivery-write | bucket-owner-read | bucket-owner-full-control) fs.s3a.multipart.purge - True if you want to purge existing multipart uploads that may not have been completed/aborted correctly (default: false) fs.s3a.multipart.purge.age - Minimum age in seconds of multipart uploads to purge (default: 86400) fs.s3a.buffer.dir - Comma separated list of directories that will be used to buffer file writes out of (default: uses ${hadoop.tmp.dir}/s3a ) *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length and MD5 to be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is still
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Steve Loughran updated HADOOP-10400: Component/s: fs/s3 Affects Version/s: 2.4.0 Incorporate new S3A FileSystem implementation - Key: HADOOP-10400 URL: https://issues.apache.org/jira/browse/HADOOP-10400 Project: Hadoop Common Issue Type: Improvement Components: fs, fs/s3 Affects Versions: 2.4.0 Reporter: Jordan Mendelson Assignee: Jordan Mendelson Attachments: HADOOP-10400-1.patch, HADOOP-10400-2.patch, HADOOP-10400-3.patch, HADOOP-10400-4.patch, HADOOP-10400-5.patch The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Tunable parameters:* fs.s3a.access.key - Your AWS access key ID (omit for role authentication) fs.s3a.secret.key - Your AWS secret key (omit for role authentication) fs.s3a.connection.maximum - Controls how many parallel connections HttpClient spawns (default: 15) fs.s3a.connection.ssl.enabled - Enables or disables SSL connections to S3 (default: true) fs.s3a.attempts.maximum - How many times we should retry commands on transient errors (default: 10) fs.s3a.connection.timeout - Socket connect timeout (default: 5000) fs.s3a.paging.maximum - How many keys to request from S3 when doing directory listings at a time (default: 5000) fs.s3a.multipart.size - How big (in bytes) to split a upload or copy operation up into (default: 104857600) fs.s3a.multipart.threshold - Until a file is this large (in bytes), use non-parallel upload (default: 2147483647) fs.s3a.acl.default - Set a canned ACL on newly created/copied objects (private | public-read | public-read-write | authenticated-read | log-delivery-write | bucket-owner-read | bucket-owner-full-control) fs.s3a.multipart.purge - True if you want to purge existing multipart uploads that may not have been completed/aborted correctly (default: false) fs.s3a.multipart.purge.age - Minimum age in seconds of multipart uploads to purge (default: 86400) fs.s3a.buffer.dir - Comma separated list of directories that will be used to buffer file writes out of (default: uses ${hadoop.tmp.dir}/s3a ) *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length and MD5 to be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is still being made for rename operations, so your job may time out unless you increase the task timeout. This driver will fully ignore _$folder$ files. This was necessary so that it could interoperate with repositories that have had the s3native driver used on them, but means that it won't recognize empty directories that s3native has been used on. Statistics for the filesystem may be calculated differently than the s3native filesystem. When uploading a file, we do not count writing the temporary file on the
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jordan Mendelson updated HADOOP-10400: -- Attachment: HADOOP-10400-5.patch This new version (-5) adjusts the test to use FileSystemContractBaseTest. However because FileSystemContractBaseTest uses the JUnit 3 runner, we can't use assume*() functions to skip the tests if we don't have a valid URL to test against, so the test has been renamed from TestS3AFileSystem to S3AFileSystemContractBaseTest and is consequently no longer run by default. A better solution would be to update FileSystemContractBaseTest for JUnit 4, but there are a lot of dependencies and that seems outside the scope of this patch. I'm not entirely sure why the linter is complaining about these patches. The last one complained about code completely outside of mine. Odd. Incorporate new S3A FileSystem implementation - Key: HADOOP-10400 URL: https://issues.apache.org/jira/browse/HADOOP-10400 Project: Hadoop Common Issue Type: Improvement Components: fs Reporter: Jordan Mendelson Assignee: Jordan Mendelson Attachments: HADOOP-10400-1.patch, HADOOP-10400-2.patch, HADOOP-10400-3.patch, HADOOP-10400-4.patch, HADOOP-10400-5.patch The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Tunable parameters:* fs.s3a.access.key - Your AWS access key ID (omit for role authentication) fs.s3a.secret.key - Your AWS secret key (omit for role authentication) fs.s3a.connection.maximum - Controls how many parallel connections HttpClient spawns (default: 15) fs.s3a.connection.ssl.enabled - Enables or disables SSL connections to S3 (default: true) fs.s3a.attempts.maximum - How many times we should retry commands on transient errors (default: 10) fs.s3a.connection.timeout - Socket connect timeout (default: 5000) fs.s3a.paging.maximum - How many keys to request from S3 when doing directory listings at a time (default: 5000) fs.s3a.multipart.size - How big (in bytes) to split a upload or copy operation up into (default: 104857600) fs.s3a.multipart.threshold - Until a file is this large (in bytes), use non-parallel upload (default: 2147483647) fs.s3a.acl.default - Set a canned ACL on newly created/copied objects (private | public-read | public-read-write | authenticated-read | log-delivery-write | bucket-owner-read | bucket-owner-full-control) fs.s3a.multipart.purge - True if you want to purge existing multipart uploads that may not have been completed/aborted correctly (default: false) fs.s3a.multipart.purge.age - Minimum age in seconds of multipart uploads to purge (default: 86400) fs.s3a.buffer.dir - Comma separated list of directories that will be used to buffer file writes out of (default: uses ${hadoop.tmp.dir}/s3a ) *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length and MD5 to be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jordan Mendelson updated HADOOP-10400: -- Attachment: HADOOP-10400-4.patch Incorporate new S3A FileSystem implementation - Key: HADOOP-10400 URL: https://issues.apache.org/jira/browse/HADOOP-10400 Project: Hadoop Common Issue Type: Improvement Components: fs Reporter: Jordan Mendelson Attachments: HADOOP-10400-1.patch, HADOOP-10400-2.patch, HADOOP-10400-3.patch, HADOOP-10400-4.patch The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Tunable parameters:* fs.s3a.access.key - Your AWS access key ID (omit for role authentication) fs.s3a.secret.key - Your AWS secret key (omit for role authentication) fs.s3a.connection.maximum - Controls how many parallel connections HttpClient spawns (default: 15) fs.s3a.connection.ssl.enabled - Enables or disables SSL connections to S3 (default: true) fs.s3a.attempts.maximum - How many times we should retry commands on transient errors (default: 10) fs.s3a.connection.timeout - Socket connect timeout (default: 5000) fs.s3a.paging.maximum - How many keys to request from S3 when doing directory listings at a time (default: 5000) fs.s3a.multipart.size - How big (in bytes) to split a upload or copy operation up into (default: 104857600) fs.s3a.multipart.threshold - Until a file is this large (in bytes), use non-parallel upload (default: 2147483647) fs.s3a.acl.default - Set a canned ACL on newly created/copied objects (private | public-read | public-read-write | authenticated-read | log-delivery-write | bucket-owner-read | bucket-owner-full-control) fs.s3a.multipart.purge - True if you want to purge existing multipart uploads that may not have been completed/aborted correctly (default: false) fs.s3a.multipart.purge.age - Minimum age in seconds of multipart uploads to purge (default: 86400) fs.s3a.buffer.dir - Comma separated list of directories that will be used to buffer file writes out of (default: uses ${hadoop.tmp.dir}/s3a ) *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length and MD5 to be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is still being made for rename operations, so your job may time out unless you increase the task timeout. This driver will fully ignore _$folder$ files. This was necessary so that it could interoperate with repositories that have had the s3native driver used on them, but means that it won't recognize empty directories that s3native has been used on. Statistics for the filesystem may be calculated differently than the s3native filesystem. When uploading a file, we do not count writing the temporary file on the local filesystem towards the local filesystem's written bytes count. When renaming files, we do not count the
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jordan Mendelson updated HADOOP-10400: -- Description: The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. It should be largely compatible with s3native except that it won't recognize s3native's empty directory marker files *_$folder$ since it uses folder/ like the Amazon's S3 explorer to denote empty directories. Other caveats: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is still being made for rename operations, so your job may time out unless you increase the task timeout. Statistics for the filesystem may be calculated differently than the s3native filesystem. When uploading a file, we do not count writing the temporary file on the local filesystem towards the local filesystem's written bytes count. When renaming files, we do not count the S3-S3 copy as read or write operations. Unlike the s3native driver, we only count bytes written when we start the upload (as opposed to the write calls to the temporary local file). The driver also counts read write ops, but they are done mostly to keep from timing out on large s3 operations. This is currently implemented as a FileSystem and not a AbstractFileSystem. was: The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. It should be largely compatible with s3native except that it won't recognize s3native's empty directory marker files *_$folder$ since it uses folder/ like the Amazon's S3 explorer to denote empty directories. Incorporate new S3A FileSystem implementation - Key: HADOOP-10400 URL: https://issues.apache.org/jira/browse/HADOOP-10400 Project: Hadoop Common Issue Type: Improvement
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jordan Mendelson updated HADOOP-10400: -- Description: The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is still being made for rename operations, so your job may time out unless you increase the task timeout. This driver will fully ignore _$folder$ files. This was necessary so that it could interoperate with repositories that have had the s3native driver used on them, but means that it won't recognize empty directories that s3native has been used on. Statistics for the filesystem may be calculated differently than the s3native filesystem. When uploading a file, we do not count writing the temporary file on the local filesystem towards the local filesystem's written bytes count. When renaming files, we do not count the S3-S3 copy as read or write operations. Unlike the s3native driver, we only count bytes written when we start the upload (as opposed to the write calls to the temporary local file). The driver also counts read write ops, but they are done mostly to keep from timing out on large s3 operations. This is currently implemented as a FileSystem and not a AbstractFileSystem. was: The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. It should be largely compatible with s3native except that it won't recognize s3native's empty directory marker files *_$folder$ since it uses folder/ like the Amazon's S3 explorer to denote empty directories. Other caveats: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jordan Mendelson updated HADOOP-10400: -- Target Version/s: (was: 2.4.0) Status: Patch Available (was: Open) Incorporate new S3A FileSystem implementation - Key: HADOOP-10400 URL: https://issues.apache.org/jira/browse/HADOOP-10400 Project: Hadoop Common Issue Type: Improvement Components: fs Reporter: Jordan Mendelson The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is still being made for rename operations, so your job may time out unless you increase the task timeout. This driver will fully ignore _$folder$ files. This was necessary so that it could interoperate with repositories that have had the s3native driver used on them, but means that it won't recognize empty directories that s3native has been used on. Statistics for the filesystem may be calculated differently than the s3native filesystem. When uploading a file, we do not count writing the temporary file on the local filesystem towards the local filesystem's written bytes count. When renaming files, we do not count the S3-S3 copy as read or write operations. Unlike the s3native driver, we only count bytes written when we start the upload (as opposed to the write calls to the temporary local file). The driver also counts read write ops, but they are done mostly to keep from timing out on large s3 operations. This is currently implemented as a FileSystem and not a AbstractFileSystem. -- This message was sent by Atlassian JIRA (v6.2#6252)
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jordan Mendelson updated HADOOP-10400: -- Attachment: HADOOP-10400-1.patch Incorporate new S3A FileSystem implementation - Key: HADOOP-10400 URL: https://issues.apache.org/jira/browse/HADOOP-10400 Project: Hadoop Common Issue Type: Improvement Components: fs Reporter: Jordan Mendelson Attachments: HADOOP-10400-1.patch The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is still being made for rename operations, so your job may time out unless you increase the task timeout. This driver will fully ignore _$folder$ files. This was necessary so that it could interoperate with repositories that have had the s3native driver used on them, but means that it won't recognize empty directories that s3native has been used on. Statistics for the filesystem may be calculated differently than the s3native filesystem. When uploading a file, we do not count writing the temporary file on the local filesystem towards the local filesystem's written bytes count. When renaming files, we do not count the S3-S3 copy as read or write operations. Unlike the s3native driver, we only count bytes written when we start the upload (as opposed to the write calls to the temporary local file). The driver also counts read write ops, but they are done mostly to keep from timing out on large s3 operations. This is currently implemented as a FileSystem and not a AbstractFileSystem. -- This message was sent by Atlassian JIRA (v6.2#6252)
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jordan Mendelson updated HADOOP-10400: -- Description: The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length and MD5 to be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is still being made for rename operations, so your job may time out unless you increase the task timeout. This driver will fully ignore _$folder$ files. This was necessary so that it could interoperate with repositories that have had the s3native driver used on them, but means that it won't recognize empty directories that s3native has been used on. Statistics for the filesystem may be calculated differently than the s3native filesystem. When uploading a file, we do not count writing the temporary file on the local filesystem towards the local filesystem's written bytes count. When renaming files, we do not count the S3-S3 copy as read or write operations. Unlike the s3native driver, we only count bytes written when we start the upload (as opposed to the write calls to the temporary local file). The driver also counts read write ops, but they are done mostly to keep from timing out on large s3 operations. This is currently implemented as a FileSystem and not a AbstractFileSystem. was: The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jordan Mendelson updated HADOOP-10400: -- Description: The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Tunable parameters:* fs.s3a.access.key - Your AWS access key ID (omit for role authentication) fs.s3a.secret.key - Your AWS secret key (omit for role authentication) fs.s3a.connection.maximum - Controls how many parallel connections HttpClient spawns (default: 15) fs.s3a.connection.ssl.enabled - Enables or disables SSL connections to S3 (default: true) fs.s3a.attempts.maximum - How many times we should retry commands on transient errors (default: 10) fs.s3a.connection.timeout - Socket connect timeout (default: 5000) fs.s3a.paging.maximum - How many keys to request from S3 when doing directory listings at a time (default: 5000) fs.s3a.multipart.size - How big (in bytes) to split a upload or copy operation up into (default: 104857600) fs.s3a.multipart.threshold - Until a file is this large (in bytes), use non-parallel upload (default: 104857600) fs.s3a.acl.default - Set a canned ACL on newly created/copied objects (private | public-read | public-read-write | authenticated-read | log-delivery-write | bucket-owner-read | bucket-owner-full-control) fs.s3a.multipart.purge - True if you want to purge existing multipart uploads that may not have been completed/aborted correctly (default: false) fs.s3a.multipart.purge.age - Minimum age in seconds of multipart uploads to purge (default: 86400) fs.s3a.buffer.dir - Comma separated list of directories that will be used to buffer file writes out of (default: uses fs.s3.buffer.dir) *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length and MD5 to be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is still being made for rename operations, so your job may time out unless you increase the task timeout. This driver will fully ignore _$folder$ files. This was necessary so that it could interoperate with repositories that have had the s3native driver used on them, but means that it won't recognize empty directories that s3native has been used on. Statistics for the filesystem may be calculated differently than the s3native filesystem. When uploading a file, we do not count writing the temporary file on the local filesystem towards the local filesystem's written bytes count. When renaming files, we do not count the S3-S3 copy as read or write operations. Unlike the s3native driver, we only count bytes written when we start the upload (as opposed to the write calls to the temporary local file). The driver also counts read write ops, but they are done mostly to keep from timing out on large s3 operations. This is currently implemented as a FileSystem and not a AbstractFileSystem. was: The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jordan Mendelson updated HADOOP-10400: -- Description: The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Tunable parameters:* fs.s3a.access.key - Your AWS access key ID (omit for role authentication) fs.s3a.secret.key - Your AWS secret key (omit for role authentication) fs.s3a.connection.maximum - Controls how many parallel connections HttpClient spawns (default: 15) fs.s3a.connection.ssl.enabled - Enables or disables SSL connections to S3 (default: true) fs.s3a.attempts.maximum - How many times we should retry commands on transient errors (default: 10) fs.s3a.connection.timeout - Socket connect timeout (default: 5000) fs.s3a.paging.maximum - How many keys to request from S3 when doing directory listings at a time (default: 5000) fs.s3a.multipart.size - How big (in bytes) to split a upload or copy operation up into (default: 104857600) fs.s3a.multipart.threshold - Until a file is this large (in bytes), use non-parallel upload (default: 524288000) fs.s3a.acl.default - Set a canned ACL on newly created/copied objects (private | public-read | public-read-write | authenticated-read | log-delivery-write | bucket-owner-read | bucket-owner-full-control) fs.s3a.multipart.purge - True if you want to purge existing multipart uploads that may not have been completed/aborted correctly (default: false) fs.s3a.multipart.purge.age - Minimum age in seconds of multipart uploads to purge (default: 86400) fs.s3a.buffer.dir - Comma separated list of directories that will be used to buffer file writes out of (default: uses fs.s3.buffer.dir) *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length and MD5 to be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is still being made for rename operations, so your job may time out unless you increase the task timeout. This driver will fully ignore _$folder$ files. This was necessary so that it could interoperate with repositories that have had the s3native driver used on them, but means that it won't recognize empty directories that s3native has been used on. Statistics for the filesystem may be calculated differently than the s3native filesystem. When uploading a file, we do not count writing the temporary file on the local filesystem towards the local filesystem's written bytes count. When renaming files, we do not count the S3-S3 copy as read or write operations. Unlike the s3native driver, we only count bytes written when we start the upload (as opposed to the write calls to the temporary local file). The driver also counts read write ops, but they are done mostly to keep from timing out on large s3 operations. This is currently implemented as a FileSystem and not a AbstractFileSystem. was: The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jordan Mendelson updated HADOOP-10400: -- Description: The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Tunable parameters:* fs.s3a.access.key - Your AWS access key ID (omit for role authentication) fs.s3a.secret.key - Your AWS secret key (omit for role authentication) fs.s3a.connection.maximum - Controls how many parallel connections HttpClient spawns (default: 15) fs.s3a.connection.ssl.enabled - Enables or disables SSL connections to S3 (default: true) fs.s3a.attempts.maximum - How many times we should retry commands on transient errors (default: 10) fs.s3a.connection.timeout - Socket connect timeout (default: 5000) fs.s3a.paging.maximum - How many keys to request from S3 when doing directory listings at a time (default: 5000) fs.s3a.multipart.size - How big (in bytes) to split a upload or copy operation up into (default: 104857600) fs.s3a.multipart.threshold - Until a file is this large (in bytes), use non-parallel upload (default: 2147483647) fs.s3a.acl.default - Set a canned ACL on newly created/copied objects (private | public-read | public-read-write | authenticated-read | log-delivery-write | bucket-owner-read | bucket-owner-full-control) fs.s3a.multipart.purge - True if you want to purge existing multipart uploads that may not have been completed/aborted correctly (default: false) fs.s3a.multipart.purge.age - Minimum age in seconds of multipart uploads to purge (default: 86400) fs.s3a.buffer.dir - Comma separated list of directories that will be used to buffer file writes out of (default: uses fs.s3.buffer.dir) *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length and MD5 to be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is still being made for rename operations, so your job may time out unless you increase the task timeout. This driver will fully ignore _$folder$ files. This was necessary so that it could interoperate with repositories that have had the s3native driver used on them, but means that it won't recognize empty directories that s3native has been used on. Statistics for the filesystem may be calculated differently than the s3native filesystem. When uploading a file, we do not count writing the temporary file on the local filesystem towards the local filesystem's written bytes count. When renaming files, we do not count the S3-S3 copy as read or write operations. Unlike the s3native driver, we only count bytes written when we start the upload (as opposed to the write calls to the temporary local file). The driver also counts read write ops, but they are done mostly to keep from timing out on large s3 operations. The AWS SDK unfortunately passes the multipart threshold as an int which means fs.s3a.multipart.threshold can not be greater than 2^31-1 (2147483647). This is currently implemented as a FileSystem and not a AbstractFileSystem. was:
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jordan Mendelson updated HADOOP-10400: -- Attachment: HADOOP-10400-2.patch Sets the core-defaults.xml file to properly match the defaults in the s3a driver. Incorporate new S3A FileSystem implementation - Key: HADOOP-10400 URL: https://issues.apache.org/jira/browse/HADOOP-10400 Project: Hadoop Common Issue Type: Improvement Components: fs Reporter: Jordan Mendelson Attachments: HADOOP-10400-1.patch, HADOOP-10400-2.patch The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Tunable parameters:* fs.s3a.access.key - Your AWS access key ID (omit for role authentication) fs.s3a.secret.key - Your AWS secret key (omit for role authentication) fs.s3a.connection.maximum - Controls how many parallel connections HttpClient spawns (default: 15) fs.s3a.connection.ssl.enabled - Enables or disables SSL connections to S3 (default: true) fs.s3a.attempts.maximum - How many times we should retry commands on transient errors (default: 10) fs.s3a.connection.timeout - Socket connect timeout (default: 5000) fs.s3a.paging.maximum - How many keys to request from S3 when doing directory listings at a time (default: 5000) fs.s3a.multipart.size - How big (in bytes) to split a upload or copy operation up into (default: 104857600) fs.s3a.multipart.threshold - Until a file is this large (in bytes), use non-parallel upload (default: 2147483647) fs.s3a.acl.default - Set a canned ACL on newly created/copied objects (private | public-read | public-read-write | authenticated-read | log-delivery-write | bucket-owner-read | bucket-owner-full-control) fs.s3a.multipart.purge - True if you want to purge existing multipart uploads that may not have been completed/aborted correctly (default: false) fs.s3a.multipart.purge.age - Minimum age in seconds of multipart uploads to purge (default: 86400) fs.s3a.buffer.dir - Comma separated list of directories that will be used to buffer file writes out of (default: uses fs.s3.buffer.dir) *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length and MD5 to be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is still being made for rename operations, so your job may time out unless you increase the task timeout. This driver will fully ignore _$folder$ files. This was necessary so that it could interoperate with repositories that have had the s3native driver used on them, but means that it won't recognize empty directories that s3native has been used on. Statistics for the filesystem may be calculated differently than the s3native filesystem. When uploading a file, we do not count writing the temporary file on the local filesystem towards the local filesystem's written bytes count. When
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jordan Mendelson updated HADOOP-10400: -- Attachment: HADOOP-10400-3.patch HADOOP-10400-3 should take care of the linter problems. Incorporate new S3A FileSystem implementation - Key: HADOOP-10400 URL: https://issues.apache.org/jira/browse/HADOOP-10400 Project: Hadoop Common Issue Type: Improvement Components: fs Reporter: Jordan Mendelson Attachments: HADOOP-10400-1.patch, HADOOP-10400-2.patch, HADOOP-10400-3.patch The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Tunable parameters:* fs.s3a.access.key - Your AWS access key ID (omit for role authentication) fs.s3a.secret.key - Your AWS secret key (omit for role authentication) fs.s3a.connection.maximum - Controls how many parallel connections HttpClient spawns (default: 15) fs.s3a.connection.ssl.enabled - Enables or disables SSL connections to S3 (default: true) fs.s3a.attempts.maximum - How many times we should retry commands on transient errors (default: 10) fs.s3a.connection.timeout - Socket connect timeout (default: 5000) fs.s3a.paging.maximum - How many keys to request from S3 when doing directory listings at a time (default: 5000) fs.s3a.multipart.size - How big (in bytes) to split a upload or copy operation up into (default: 104857600) fs.s3a.multipart.threshold - Until a file is this large (in bytes), use non-parallel upload (default: 2147483647) fs.s3a.acl.default - Set a canned ACL on newly created/copied objects (private | public-read | public-read-write | authenticated-read | log-delivery-write | bucket-owner-read | bucket-owner-full-control) fs.s3a.multipart.purge - True if you want to purge existing multipart uploads that may not have been completed/aborted correctly (default: false) fs.s3a.multipart.purge.age - Minimum age in seconds of multipart uploads to purge (default: 86400) fs.s3a.buffer.dir - Comma separated list of directories that will be used to buffer file writes out of (default: uses fs.s3.buffer.dir) *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length and MD5 to be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is still being made for rename operations, so your job may time out unless you increase the task timeout. This driver will fully ignore _$folder$ files. This was necessary so that it could interoperate with repositories that have had the s3native driver used on them, but means that it won't recognize empty directories that s3native has been used on. Statistics for the filesystem may be calculated differently than the s3native filesystem. When uploading a file, we do not count writing the temporary file on the local filesystem towards the local filesystem's written bytes count. When
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jordan Mendelson updated HADOOP-10400: -- Description: The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Tunable parameters:* fs.s3a.access.key - Your AWS access key ID (omit for role authentication) fs.s3a.secret.key - Your AWS secret key (omit for role authentication) fs.s3a.connection.maximum - Controls how many parallel connections HttpClient spawns (default: 15) fs.s3a.connection.ssl.enabled - Enables or disables SSL connections to S3 (default: true) fs.s3a.attempts.maximum - How many times we should retry commands on transient errors (default: 10) fs.s3a.connection.timeout - Socket connect timeout (default: 5000) fs.s3a.paging.maximum - How many keys to request from S3 when doing directory listings at a time (default: 5000) fs.s3a.multipart.size - How big (in bytes) to split a upload or copy operation up into (default: 104857600) fs.s3a.multipart.threshold - Until a file is this large (in bytes), use non-parallel upload (default: 2147483647) fs.s3a.acl.default - Set a canned ACL on newly created/copied objects (private | public-read | public-read-write | authenticated-read | log-delivery-write | bucket-owner-read | bucket-owner-full-control) fs.s3a.multipart.purge - True if you want to purge existing multipart uploads that may not have been completed/aborted correctly (default: false) fs.s3a.multipart.purge.age - Minimum age in seconds of multipart uploads to purge (default: 86400) fs.s3a.buffer.dir - Comma separated list of directories that will be used to buffer file writes out of (default: uses ${hadoop.tmp.dir}/s3a) *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length and MD5 to be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is still being made for rename operations, so your job may time out unless you increase the task timeout. This driver will fully ignore _$folder$ files. This was necessary so that it could interoperate with repositories that have had the s3native driver used on them, but means that it won't recognize empty directories that s3native has been used on. Statistics for the filesystem may be calculated differently than the s3native filesystem. When uploading a file, we do not count writing the temporary file on the local filesystem towards the local filesystem's written bytes count. When renaming files, we do not count the S3-S3 copy as read or write operations. Unlike the s3native driver, we only count bytes written when we start the upload (as opposed to the write calls to the temporary local file). The driver also counts read write ops, but they are done mostly to keep from timing out on large s3 operations. The AWS SDK unfortunately passes the multipart threshold as an int which means fs.s3a.multipart.threshold can not be greater than 2^31-1 (2147483647). This is currently implemented as a FileSystem and not a AbstractFileSystem.
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jordan Mendelson updated HADOOP-10400: -- Description: The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Tunable parameters:* fs.s3a.access.key - Your AWS access key ID (omit for role authentication) fs.s3a.secret.key - Your AWS secret key (omit for role authentication) fs.s3a.connection.maximum - Controls how many parallel connections HttpClient spawns (default: 15) fs.s3a.connection.ssl.enabled - Enables or disables SSL connections to S3 (default: true) fs.s3a.attempts.maximum - How many times we should retry commands on transient errors (default: 10) fs.s3a.connection.timeout - Socket connect timeout (default: 5000) fs.s3a.paging.maximum - How many keys to request from S3 when doing directory listings at a time (default: 5000) fs.s3a.multipart.size - How big (in bytes) to split a upload or copy operation up into (default: 104857600) fs.s3a.multipart.threshold - Until a file is this large (in bytes), use non-parallel upload (default: 2147483647) fs.s3a.acl.default - Set a canned ACL on newly created/copied objects (private | public-read | public-read-write | authenticated-read | log-delivery-write | bucket-owner-read | bucket-owner-full-control) fs.s3a.multipart.purge - True if you want to purge existing multipart uploads that may not have been completed/aborted correctly (default: false) fs.s3a.multipart.purge.age - Minimum age in seconds of multipart uploads to purge (default: 86400) fs.s3a.buffer.dir - Comma separated list of directories that will be used to buffer file writes out of (default: uses ${hadoop.tmp.dir}/s3a ) *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length and MD5 to be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is still being made for rename operations, so your job may time out unless you increase the task timeout. This driver will fully ignore _$folder$ files. This was necessary so that it could interoperate with repositories that have had the s3native driver used on them, but means that it won't recognize empty directories that s3native has been used on. Statistics for the filesystem may be calculated differently than the s3native filesystem. When uploading a file, we do not count writing the temporary file on the local filesystem towards the local filesystem's written bytes count. When renaming files, we do not count the S3-S3 copy as read or write operations. Unlike the s3native driver, we only count bytes written when we start the upload (as opposed to the write calls to the temporary local file). The driver also counts read write ops, but they are done mostly to keep from timing out on large s3 operations. The AWS SDK unfortunately passes the multipart threshold as an int which means fs.s3a.multipart.threshold can not be greater than 2^31-1 (2147483647). This is currently implemented as a FileSystem and not a AbstractFileSystem.
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jordan Mendelson updated HADOOP-10400: -- Status: Open (was: Patch Available) Incorporate new S3A FileSystem implementation - Key: HADOOP-10400 URL: https://issues.apache.org/jira/browse/HADOOP-10400 Project: Hadoop Common Issue Type: Improvement Components: fs Reporter: Jordan Mendelson Attachments: HADOOP-10400-1.patch, HADOOP-10400-2.patch, HADOOP-10400-3.patch The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Tunable parameters:* fs.s3a.access.key - Your AWS access key ID (omit for role authentication) fs.s3a.secret.key - Your AWS secret key (omit for role authentication) fs.s3a.connection.maximum - Controls how many parallel connections HttpClient spawns (default: 15) fs.s3a.connection.ssl.enabled - Enables or disables SSL connections to S3 (default: true) fs.s3a.attempts.maximum - How many times we should retry commands on transient errors (default: 10) fs.s3a.connection.timeout - Socket connect timeout (default: 5000) fs.s3a.paging.maximum - How many keys to request from S3 when doing directory listings at a time (default: 5000) fs.s3a.multipart.size - How big (in bytes) to split a upload or copy operation up into (default: 104857600) fs.s3a.multipart.threshold - Until a file is this large (in bytes), use non-parallel upload (default: 2147483647) fs.s3a.acl.default - Set a canned ACL on newly created/copied objects (private | public-read | public-read-write | authenticated-read | log-delivery-write | bucket-owner-read | bucket-owner-full-control) fs.s3a.multipart.purge - True if you want to purge existing multipart uploads that may not have been completed/aborted correctly (default: false) fs.s3a.multipart.purge.age - Minimum age in seconds of multipart uploads to purge (default: 86400) fs.s3a.buffer.dir - Comma separated list of directories that will be used to buffer file writes out of (default: uses ${hadoop.tmp.dir}/s3a ) *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length and MD5 to be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is still being made for rename operations, so your job may time out unless you increase the task timeout. This driver will fully ignore _$folder$ files. This was necessary so that it could interoperate with repositories that have had the s3native driver used on them, but means that it won't recognize empty directories that s3native has been used on. Statistics for the filesystem may be calculated differently than the s3native filesystem. When uploading a file, we do not count writing the temporary file on the local filesystem towards the local filesystem's written bytes count. When renaming files, we do not count the S3-S3 copy as read
[jira] [Updated] (HADOOP-10400) Incorporate new S3A FileSystem implementation
[ https://issues.apache.org/jira/browse/HADOOP-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jordan Mendelson updated HADOOP-10400: -- Status: Patch Available (was: Open) Incorporate new S3A FileSystem implementation - Key: HADOOP-10400 URL: https://issues.apache.org/jira/browse/HADOOP-10400 Project: Hadoop Common Issue Type: Improvement Components: fs Reporter: Jordan Mendelson Attachments: HADOOP-10400-1.patch, HADOOP-10400-2.patch, HADOOP-10400-3.patch The s3native filesystem has a number of limitations (some of which were recently fixed by HADOOP-9454). This patch adds an s3a filesystem which uses the aws-sdk instead of the jets3t library. There are a number of improvements over s3native including: - Parallel copy (rename) support (dramatically speeds up commits on large files) - AWS S3 explorer compatible empty directories files xyz/ instead of xyz_$folder$ (reduces littering) - Ignores s3native created _$folder$ files created by s3native and other S3 browsing utilities - Supports multiple output buffer dirs to even out IO when uploading files - Supports IAM role-based authentication - Allows setting a default canned ACL for uploads (public, private, etc.) - Better error recovery handling - Should handle input seeks without having to download the whole file (used for splits a lot) This code is a copy of https://github.com/Aloisius/hadoop-s3a with patches to various pom files to get it to build against trunk. I've been using 0.0.1 in production with CDH 4 for several months and CDH 5 for a few days. The version here is 0.0.2 which changes around some keys to hopefully bring the key name style more inline with the rest of hadoop 2.x. *Tunable parameters:* fs.s3a.access.key - Your AWS access key ID (omit for role authentication) fs.s3a.secret.key - Your AWS secret key (omit for role authentication) fs.s3a.connection.maximum - Controls how many parallel connections HttpClient spawns (default: 15) fs.s3a.connection.ssl.enabled - Enables or disables SSL connections to S3 (default: true) fs.s3a.attempts.maximum - How many times we should retry commands on transient errors (default: 10) fs.s3a.connection.timeout - Socket connect timeout (default: 5000) fs.s3a.paging.maximum - How many keys to request from S3 when doing directory listings at a time (default: 5000) fs.s3a.multipart.size - How big (in bytes) to split a upload or copy operation up into (default: 104857600) fs.s3a.multipart.threshold - Until a file is this large (in bytes), use non-parallel upload (default: 2147483647) fs.s3a.acl.default - Set a canned ACL on newly created/copied objects (private | public-read | public-read-write | authenticated-read | log-delivery-write | bucket-owner-read | bucket-owner-full-control) fs.s3a.multipart.purge - True if you want to purge existing multipart uploads that may not have been completed/aborted correctly (default: false) fs.s3a.multipart.purge.age - Minimum age in seconds of multipart uploads to purge (default: 86400) fs.s3a.buffer.dir - Comma separated list of directories that will be used to buffer file writes out of (default: uses ${hadoop.tmp.dir}/s3a ) *Caveats*: Hadoop uses a standard output committer which uploads files as filename.COPYING before renaming them. This can cause unnecessary performance issues with S3 because it does not have a rename operation and S3 already verifies uploads against an md5 that the driver sets on the upload request. While this FileSystem should be significantly faster than the built-in s3native driver because of parallel copy support, you may want to consider setting a null output committer on our jobs to further improve performance. Because S3 requires the file length and MD5 to be known before a file is uploaded, all output is buffered out to a temporary file first similar to the s3native driver. Due to the lack of native rename() for S3, renaming extremely large files or directories make take a while. Unfortunately, there is no way to notify hadoop that progress is still being made for rename operations, so your job may time out unless you increase the task timeout. This driver will fully ignore _$folder$ files. This was necessary so that it could interoperate with repositories that have had the s3native driver used on them, but means that it won't recognize empty directories that s3native has been used on. Statistics for the filesystem may be calculated differently than the s3native filesystem. When uploading a file, we do not count writing the temporary file on the local filesystem towards the local filesystem's written bytes count. When renaming files, we do not count the S3-S3 copy as read