You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ozone.apache.org by GitBox <gi...@apache.org> on 2020/04/16 11:40:51 UTC
[GitHub] [hadoop-ozone] sadanand48 opened a new pull request #833:
HDDS-3314. scmcli container info command failing intermittently
sadanand48 opened a new pull request #833: HDDS-3314. scmcli container info command failing intermittently
URL: https://github.com/apache/hadoop-ozone/pull/833
## What changes were proposed in this pull request?
scmcli info command fails intermittently on killing a datanode in a multinode setup as it uses the ratis pipeline.This was because the query happens only on the leader node.The change is that the scm now queries different datanodes in case of failure through the grpcClient.
## What is the link to the Apache JIRA
https://issues.apache.org/jira/browse/HDDS-3314
## How was this patch tested?
Manual test
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: ozone-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: ozone-issues-help@hadoop.apache.org
[GitHub] [hadoop-ozone] bharatviswa504 commented on a change in pull request
#833: HDDS-3314. scmcli container info command failing intermittently
Posted by GitBox <gi...@apache.org>.
bharatviswa504 commented on a change in pull request #833: HDDS-3314. scmcli container info command failing intermittently
URL: https://github.com/apache/hadoop-ozone/pull/833#discussion_r409885190
##########
File path: hadoop-hdds/tools/src/main/java/org/apache/hadoop/hdds/scm/cli/ContainerOperationClient.java
##########
@@ -382,7 +382,7 @@ public ContainerDataProto readContainer(long containerID,
Pipeline pipeline) throws IOException {
XceiverClientSpi client = null;
try {
- client = xceiverClientManager.acquireClient(pipeline);
+ client = xceiverClientManager.acquireClientForReadData(pipeline);
Review comment:
I have not really understood this change, as we just changed from acquireClient -> acquireClientForReadData.
How that will handle failures, and how it will retry on other datanodes.
And also InfoSubCommand Uses below code, and it is not calling readContainer.
final ContainerWithPipeline container = scmClient.
getContainerWithPipeline(containerID);
> final ContainerWithPipeline container = scmClient.
> getContainerWithPipeline(containerID);
> Preconditions.checkNotNull(container, "Container cannot be null");
If possible, can you explain this change, not sure if I am missing something here.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: ozone-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: ozone-issues-help@hadoop.apache.org
[GitHub] [hadoop-ozone] bshashikant commented on a change in pull request
#833: HDDS-3314. FIx ContainerOperationClient#readContainer to use Grpc
Client to read from datanode
Posted by GitBox <gi...@apache.org>.
bshashikant commented on a change in pull request #833: HDDS-3314. FIx ContainerOperationClient#readContainer to use Grpc Client to read from datanode
URL: https://github.com/apache/hadoop-ozone/pull/833#discussion_r410099334
##########
File path: hadoop-hdds/tools/src/main/java/org/apache/hadoop/hdds/scm/cli/ContainerOperationClient.java
##########
@@ -382,7 +382,7 @@ public ContainerDataProto readContainer(long containerID,
Pipeline pipeline) throws IOException {
XceiverClientSpi client = null;
try {
- client = xceiverClientManager.acquireClient(pipeline);
+ client = xceiverClientManager.acquireClientForReadData(pipeline);
Review comment:
@bharat, if we use acquireClientForReadData, it will use XceiverClientGrpc to read the info rather than Ratis(which reads only from the leader if we use acquireClient). XceiverClientGrpc automatically fails over from datanode to another and retries if it sees a failure.
InfoSubCommand has recently been changed to not use readContainer call to read data from datanode. But , in Ozone, as we only use XceiverClientGrpc to read from datanode , it make sense here to use the same while reading the containerData as well (acquireClient -> acquireClientForReadData). This function is used by ChunkKeyHandler. I have modified the description of the ira accordingly.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: ozone-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: ozone-issues-help@hadoop.apache.org
[GitHub] [hadoop-ozone] bshashikant commented on a change in pull request
#833: HDDS-3314. scmcli container info command failing intermittently
Posted by GitBox <gi...@apache.org>.
bshashikant commented on a change in pull request #833: HDDS-3314. scmcli container info command failing intermittently
URL: https://github.com/apache/hadoop-ozone/pull/833#discussion_r410099334
##########
File path: hadoop-hdds/tools/src/main/java/org/apache/hadoop/hdds/scm/cli/ContainerOperationClient.java
##########
@@ -382,7 +382,7 @@ public ContainerDataProto readContainer(long containerID,
Pipeline pipeline) throws IOException {
XceiverClientSpi client = null;
try {
- client = xceiverClientManager.acquireClient(pipeline);
+ client = xceiverClientManager.acquireClientForReadData(pipeline);
Review comment:
@bharat, if we use acquireClientForReadData, it will use XceiverClientGrpc to read the info rather than Ratis(which reads from the leader). XceiverClientGrpc automatically fails over from datanode to another if it sees a failure.
InfoSubCommand has recently been changed to not use readContainer call to read data from datanode. But , in Ozone, we only use XceiverClientGrpc to read from datanode , it make sense here to use the same while reading the containerData as well (acquireClient -> acquireClientForReadData). This function is used by ChunkKeyHandler.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: ozone-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: ozone-issues-help@hadoop.apache.org
[GitHub] [hadoop-ozone] bshashikant commented on a change in pull request
#833: HDDS-3314. scmcli container info command failing intermittently
Posted by GitBox <gi...@apache.org>.
bshashikant commented on a change in pull request #833: HDDS-3314. scmcli container info command failing intermittently
URL: https://github.com/apache/hadoop-ozone/pull/833#discussion_r410099334
##########
File path: hadoop-hdds/tools/src/main/java/org/apache/hadoop/hdds/scm/cli/ContainerOperationClient.java
##########
@@ -382,7 +382,7 @@ public ContainerDataProto readContainer(long containerID,
Pipeline pipeline) throws IOException {
XceiverClientSpi client = null;
try {
- client = xceiverClientManager.acquireClient(pipeline);
+ client = xceiverClientManager.acquireClientForReadData(pipeline);
Review comment:
@bharat, if we use acquireClientForReadData, it will use XceiverClientGrpc to read the info rather than Ratis(which reads only from the leader if we use acquireClient). XceiverClientGrpc automatically fails over from datanode to another and retries if it sees a failure.
InfoSubCommand has recently been changed to not use readContainer call to read data from datanode. But , in Ozone, as we only use XceiverClientGrpc to read from datanode , it make sense here to use the same while reading the containerData as well (acquireClient -> acquireClientForReadData). This function is used by ChunkKeyHandler.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: ozone-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: ozone-issues-help@hadoop.apache.org
[GitHub] [hadoop-ozone] bharatviswa504 commented on a change in pull request
#833: HDDS-3314. scmcli container info command failing intermittently
Posted by GitBox <gi...@apache.org>.
bharatviswa504 commented on a change in pull request #833: HDDS-3314. scmcli container info command failing intermittently
URL: https://github.com/apache/hadoop-ozone/pull/833#discussion_r409885190
##########
File path: hadoop-hdds/tools/src/main/java/org/apache/hadoop/hdds/scm/cli/ContainerOperationClient.java
##########
@@ -382,7 +382,7 @@ public ContainerDataProto readContainer(long containerID,
Pipeline pipeline) throws IOException {
XceiverClientSpi client = null;
try {
- client = xceiverClientManager.acquireClient(pipeline);
+ client = xceiverClientManager.acquireClientForReadData(pipeline);
Review comment:
I have not really understood this change, as we just changed from acquireClient -> acquireClientForReadData.
How that will handle failures, and how it will retry on other datanodes.
And also InfoSubCommand Uses below code, and it is not calling readContainer.
final ContainerWithPipeline container = scmClient.
getContainerWithPipeline(containerID);
> final ContainerWithPipeline container = scmClient.
> getContainerWithPipeline(containerID);
> Preconditions.checkNotNull(container, "Container cannot be null");
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: ozone-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: ozone-issues-help@hadoop.apache.org