[jira] [Comment Edited] (HDDS-155) Implement KeyValueContainer and adopt new disk layout for the containers
[ https://issues.apache.org/jira/browse/HDDS-155?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16513300#comment-16513300 ] Bharat Viswanadham edited comment on HDDS-155 at 6/15/18 3:57 AM: -- I will commit this shortly. shaded-client issue is not caused by this patch. was (Author: bharatviswa): I will commit this shortly. shaded-client issue is not caused by this patch. > Implement KeyValueContainer and adopt new disk layout for the containers > > > Key: HDDS-155 > URL: https://issues.apache.org/jira/browse/HDDS-155 > Project: Hadoop Distributed Data Store > Issue Type: Sub-task >Reporter: Bharat Viswanadham >Assignee: Bharat Viswanadham >Priority: Major > Fix For: 0.2.1 > > Attachments: HDDS-155-HDDS-48.00.patch, HDDS-155-HDDS-48.01.patch, > HDDS-155-HDDS-48.02.patch, HDDS-155-HDDS-48.03.patch, > HDDS-155-HDDS-48.04.patch, HDDS-155-HDDS-48.05.patch, > HDDS-155-HDDS-48.06.patch, HDDS-155-HDDS-48.07.patch, > HDDS-155-HDDS-48.08.patch > > > This Jira is to add following: > # Implement Container Interface > # Use new directory layout proposed in the design document. > a. Data location (chunks) > b. Meta location (DB and .container files) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
[jira] [Comment Edited] (HDDS-155) Implement KeyValueContainer and adopt new disk layout for the containers
[ https://issues.apache.org/jira/browse/HDDS-155?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16513133#comment-16513133 ] Bharat Viswanadham edited comment on HDDS-155 at 6/14/18 11:32 PM: --- Hi [~hanishakoneru] Thanks for the review. Addressed review comments and Jenkins reported issues in patch v06. was (Author: bharatviswa): Hi [~hanishakoneru] Thanks for the review. Addressed review comments in patch v06. > Implement KeyValueContainer and adopt new disk layout for the containers > > > Key: HDDS-155 > URL: https://issues.apache.org/jira/browse/HDDS-155 > Project: Hadoop Distributed Data Store > Issue Type: Sub-task >Reporter: Bharat Viswanadham >Assignee: Bharat Viswanadham >Priority: Major > Fix For: 0.2.1 > > Attachments: HDDS-155-HDDS-48.00.patch, HDDS-155-HDDS-48.01.patch, > HDDS-155-HDDS-48.02.patch, HDDS-155-HDDS-48.03.patch, > HDDS-155-HDDS-48.04.patch, HDDS-155-HDDS-48.05.patch, > HDDS-155-HDDS-48.06.patch > > > This Jira is to add following: > # Implement Container Interface > # Use new directory layout proposed in the design document. > a. Data location (chunks) > b. Meta location (DB and .container files) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
[jira] [Comment Edited] (HDDS-155) Implement KeyValueContainer and adopt new disk layout for the containers
[ https://issues.apache.org/jira/browse/HDDS-155?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16513041#comment-16513041 ] Hanisha Koneru edited comment on HDDS-155 at 6/14/18 9:59 PM: -- Thanks for the update [~bharatviswa]. Some comments (Many of them are just NITs.): # Can we rename the following in {{DatanodeContainerProtocol.proto}} CONTAINER_REQUIRED_FILES_CREATE_ERROR -> CONTAINER_FILES_CREATE_ERROR CONTAINER_CHECKSUM_FILE_CALCULATE_ERROR -> CONTAINER_CHECKSUM_ERROR # {{KeyValueContainer}} Line 131, 133 : Instead of "String.valueOf(containerId)", we can reuse containerName # {{KeyValueContainer#updateRequiredFiles()}} -> when update succeeds, the backup files must be deleted. # In all scenarios where the createContainer fails, we should delete the containerBase dir. # Can we rename {{createRequiredFiles}} and {{updateRequiredFiles}} to createContainerFile and updateContainerFile? Required doesn’t tell us instantly what the files are. I think it is ok not to mention that the checksum file is also created as part of createContainerFile. # Line 225 : "Unable to delete container checksum backup file” -> Since we are deleting the temporary file, we can have "Unable to delete container temporary checksum file”. # {{KeyValueContainer#updateRequiredFiles()}}, when throwing INVALID_CONTAINER_STATE or when the restore fails, we should set the state of the container to INVALID. # When restoring from back up files, can we add a info log message saying "update failed, so restoring the container files." # Javadoc of computeChecksum is misleading. "Create checksum file of the .container file.” -> “Computer checksum of the .container file”. # Can we have a file handle to {{.db file}} in {{KeyValueContainerData}} as this would be used by every key operation. # {{KeyValueContainerUtil#verifyIsNewContainer()}} -> existence of containerBasePath should be sufficient condition to fail createContainer. was (Author: hanishakoneru): Thanks for the update [~bharatviswa]. Some comments (Many of them are just NITs.): # Can we rename the following in {{DatanodeContainerProtocol.proto}} CONTAINER_REQUIRED_FILES_CREATE_ERROR -> CONTAINER_FILES_CREATE_ERROR CONTAINER_CHECKSUM_FILE_CALCULATE_ERROR -> CONTAINER_CHECKSUM_ERROR # {{KeyValueContainer}} Line 131, 133 : Instead of "String.valueOf(containerId)", we can reuse containerName # {{KeyValueContainer#updateRequiredFiles()}} -> when update succeeds, the backup files must be deleted. In all scenarios where the createContainer fails, we should delete the containerBase dir. # Can we rename {{createRequiredFiles}} and {{updateRequiredFiles}} to createContainerFile and updateContainerFile? Required doesn’t tell us instantly what the files are. I think it is ok not to mention that the checksum file is also created as part of createContainerFile. # Line 225 : "Unable to delete container checksum backup file” -> Since we are deleting the temporary file, we can have "Unable to delete container temporary checksum file”. # {{KeyValueContainer#updateRequiredFiles()}}, when throwing INVALID_CONTAINER_STATE or when the restore fails, we should set the state of the container to INVALID. # When restoring from back up files, can we add a info log message saying "update failed, so restoring the container files." # Javadoc of computeChecksum is misleading. "Create checksum file of the .container file.” -> “Computer checksum of the .container file”. # Can we have a file handle to {{.db file}} in {{KeyValueContainerData}} as this would be used by every key operation. # {{KeyValueContainerUtil#verifyIsNewContainer()}} -> existence of containerBasePath should be sufficient condition to fail createContainer. > Implement KeyValueContainer and adopt new disk layout for the containers > > > Key: HDDS-155 > URL: https://issues.apache.org/jira/browse/HDDS-155 > Project: Hadoop Distributed Data Store > Issue Type: Sub-task >Reporter: Bharat Viswanadham >Assignee: Bharat Viswanadham >Priority: Major > Fix For: 0.2.1 > > Attachments: HDDS-155-HDDS-48.00.patch, HDDS-155-HDDS-48.01.patch, > HDDS-155-HDDS-48.02.patch, HDDS-155-HDDS-48.03.patch, > HDDS-155-HDDS-48.04.patch, HDDS-155-HDDS-48.05.patch > > > This Jira is to add following: > # Implement Container Interface > # Use new directory layout proposed in the design document. > a. Data location (chunks) > b. Meta location (DB and .container files) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
[jira] [Comment Edited] (HDDS-155) Implement KeyValueContainer and adopt new disk layout for the containers
[ https://issues.apache.org/jira/browse/HDDS-155?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16513041#comment-16513041 ] Hanisha Koneru edited comment on HDDS-155 at 6/14/18 9:59 PM: -- Thanks for the update [~bharatviswa]. Some comments (Many of them are just NITs.): # Can we rename the following in {{DatanodeContainerProtocol.proto}} CONTAINER_REQUIRED_FILES_CREATE_ERROR -> CONTAINER_FILES_CREATE_ERROR CONTAINER_CHECKSUM_FILE_CALCULATE_ERROR -> CONTAINER_CHECKSUM_ERROR # {{KeyValueContainer}} Line 131, 133 : Instead of "String.valueOf(containerId)", we can reuse containerName # {{KeyValueContainer#updateRequiredFiles()}} -> when update succeeds, the backup files must be deleted. In all scenarios where the createContainer fails, we should delete the containerBase dir. # Can we rename {{createRequiredFiles}} and {{updateRequiredFiles}} to createContainerFile and updateContainerFile? Required doesn’t tell us instantly what the files are. I think it is ok not to mention that the checksum file is also created as part of createContainerFile. # Line 225 : "Unable to delete container checksum backup file” -> Since we are deleting the temporary file, we can have "Unable to delete container temporary checksum file”. # {{KeyValueContainer#updateRequiredFiles()}}, when throwing INVALID_CONTAINER_STATE or when the restore fails, we should set the state of the container to INVALID. # When restoring from back up files, can we add a info log message saying "update failed, so restoring the container files." # Javadoc of computeChecksum is misleading. "Create checksum file of the .container file.” -> “Computer checksum of the .container file”. # Can we have a file handle to {{.db file}} in {{KeyValueContainerData}} as this would be used by every key operation. # {{KeyValueContainerUtil#verifyIsNewContainer()}} -> existence of containerBasePath should be sufficient condition to fail createContainer. was (Author: hanishakoneru): Thanks for the update [~bharatviswa]. Some comments (Many of them are just NITs.): # Can we rename the following in {{DatanodeContainerProtocol.proto}} CONTAINER_REQUIRED_FILES_CREATE_ERROR -> CONTAINER_FILES_CREATE_ERROR CONTAINER_CHECKSUM_FILE_CALCULATE_ERROR -> CONTAINER_CHECKSUM_ERROR # {{KeyValueContainer}} Line 131, 133 : Instead of "String.valueOf(containerId)", we can reuse containerName # {{KeyValueContainer#createRequiredFiles()}} -> there is a possibility that the rename of .containerFile succeeds but .checksum file fails. In this scenario and all other scenarios where the createContainer fails, we should delete the containerBase dir. # Can we rename {{createRequiredFiles}} and {{updateRequiredFiles}} to createContainerFile and updateContainerFile? Required doesn’t tell us instantly what the files are. I think it is ok not to mention that the checksum file is also created as part of createContainerFile. # Line 225 : "Unable to delete container checksum backup file” -> Since we are deleting the temporary file, we can have "Unable to delete container temporary checksum file”. # {{KeyValueContainer#updateRequiredFiles()}}, when throwing INVALID_CONTAINER_STATE or when the restore fails, we should set the state of the container to INVALID. # When restoring from back up files, can we add a info log message saying "update failed, so restoring the container files." # Javadoc of computeChecksum is misleading. "Create checksum file of the .container file.” -> “Computer checksum of the .container file”. # Can we have a file handle to {{.db file}} in {{KeyValueContainerData}} as this would be used by every key operation. # {{KeyValueContainerUtil#verifyIsNewContainer()}} -> existence of containerBasePath should be sufficient condition to fail createContainer. > Implement KeyValueContainer and adopt new disk layout for the containers > > > Key: HDDS-155 > URL: https://issues.apache.org/jira/browse/HDDS-155 > Project: Hadoop Distributed Data Store > Issue Type: Sub-task >Reporter: Bharat Viswanadham >Assignee: Bharat Viswanadham >Priority: Major > Fix For: 0.2.1 > > Attachments: HDDS-155-HDDS-48.00.patch, HDDS-155-HDDS-48.01.patch, > HDDS-155-HDDS-48.02.patch, HDDS-155-HDDS-48.03.patch, > HDDS-155-HDDS-48.04.patch, HDDS-155-HDDS-48.05.patch > > > This Jira is to add following: > # Implement Container Interface > # Use new directory layout proposed in the design document. > a. Data location (chunks) > b. Meta location (DB and .container files) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org For additional commands,
[jira] [Comment Edited] (HDDS-155) Implement KeyValueContainer and adopt new disk layout for the containers
[ https://issues.apache.org/jira/browse/HDDS-155?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16512841#comment-16512841 ] Bharat Viswanadham edited comment on HDDS-155 at 6/14/18 6:16 PM: -- Rebased patch to use new code changes done in HDDS-156 Will submit the patch, once HDDS-156 gets committed. was (Author: bharatviswa): Rebased patch to use new code changes done in HDDS-155. Will submit the patch, once HDDS-155 gets committed. > Implement KeyValueContainer and adopt new disk layout for the containers > > > Key: HDDS-155 > URL: https://issues.apache.org/jira/browse/HDDS-155 > Project: Hadoop Distributed Data Store > Issue Type: Sub-task >Reporter: Bharat Viswanadham >Assignee: Bharat Viswanadham >Priority: Major > Fix For: 0.2.1 > > Attachments: HDDS-155-HDDS-48.00.patch, HDDS-155-HDDS-48.01.patch, > HDDS-155-HDDS-48.02.patch, HDDS-155-HDDS-48.03.patch, > HDDS-155-HDDS-48.04.patch, HDDS-155-HDDS-48.05.patch > > > This Jira is to add following: > # Implement Container Interface > # Use new directory layout proposed in the design document. > a. Data location (chunks) > b. Meta location (DB and .container files) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org