SofiaSedova commented on code in PR #12156:
URL: https://github.com/apache/ignite/pull/12156#discussion_r2176885829
##########
docs/_docs/distributed-computing/map-reduce.adoc:
##########
@@ -89,7 +89,17 @@ include::{javaFile}[tags=failover,indent=0]
== Compute Task Adapters
There are several helper classes that provide most commonly used
implementations of the `result(...)` and `map(...)` methods.
-* `ComputeTaskAdapter` — This class implements the `result()` method to return
the `FAILOVER` policy if a job throws an exception and the `WAIT` policy
otherwise. It means that this implementation will wait for all jobs to finish
with a result.
+* `ComputeTaskAdapter` — This class provides implementation for the `result()`
method, which makes the best effort to wait for all Compute Jobs to complete
successfully and process their results (see `WAIT` policy).
Review Comment:
```suggestion
* `ComputeTaskAdapter` — this class provides implementation for the
`result()` method, which makes the best effort to wait for all Compute Jobs to
complete successfully and process their results (see `WAIT` policy).
```
##########
docs/_docs/distributed-computing/map-reduce.adoc:
##########
@@ -89,7 +89,17 @@ include::{javaFile}[tags=failover,indent=0]
== Compute Task Adapters
There are several helper classes that provide most commonly used
implementations of the `result(...)` and `map(...)` methods.
-* `ComputeTaskAdapter` — This class implements the `result()` method to return
the `FAILOVER` policy if a job throws an exception and the `WAIT` policy
otherwise. It means that this implementation will wait for all jobs to finish
with a result.
+* `ComputeTaskAdapter` — This class provides implementation for the `result()`
method, which makes the best effort to wait for all Compute Jobs to complete
successfully and process their results (see `WAIT` policy).
++
+--
+Comute Jobs will be automatically redeployed (see `FAILOVER` policy) in the
following scenarious:
+
+- the user-provided executor refused to process the mapped Compute Job;
+- the node intended to run the Compute Job is no longer present in the cluster;
+- the user code of the Compute Job explicitly threw a dedicated exception -
`ComputeJobFailoverException`.
Review Comment:
```suggestion
- the user code of the Compute Job explicitly threw a dedicated exception —
`ComputeJobFailoverException`.
```
##########
docs/_docs/key-value-api/transactions.adoc:
##########
@@ -85,6 +85,11 @@
include::code-snippets/cpp/src/transactions.cpp[tag=transactions-execution,inden
--
+[WARNING]
+====
+It is critical that an Ignite Trusnsaction be `closed` regardless of its
commit state. This ensures that all resources are released and the transaction
is no longer bind to the current thread. Therefore, the `try-with-resource`
block is highly recommended when working with transactions.
Review Comment:
```suggestion
It is critical that an Ignite Transaction is `closed` regardless of its
commit state. This ensures that all resources are released and the transaction
is no longer bound to the current thread. Therefore, the `try-with-resource`
block is highly recommended when working with transactions.
```
##########
docs/_docs/perf-and-troubleshooting/handling-exceptions.adoc:
##########
@@ -246,3 +246,8 @@ include::{javaFile}[tag=failure-types,indent=0]
----
--
+=== Handling Compute Job Exceptions
+
+If
link:https://ignite.apache.org/docs/latest/distributed-computing/map-reduce#:~:text=and%20map(%E2%80%A6%E2%80%8B)%20methods.-,ComputeTaskAdapter,-%E2%80%94%20This%20class%20implements[ComputeTaskAdapter]
is used as Compute Task implementation and any of mapped Compute Jobs thrown
exception during execution, the entire Compute Task immediately fails with an
`IgniteException`, and all remaining Compute Jobs are stopped asynchronously.
Compute Job failures can be tracked in the Ignite log for the message `Remote
job throws user exception…`.
Review Comment:
```suggestion
If
link:https://ignite.apache.org/docs/latest/distributed-computing/map-reduce#:~:text=and%20map(%E2%80%A6%E2%80%8B)%20methods.-,ComputeTaskAdapter,-%E2%80%94%20This%20class%20implements[ComputeTaskAdapter]
is used as Compute Task implementation and any of mapped Compute Jobs throws
an exception during execution, the entire Compute Task immediately fails with
an `IgniteException`, and all remaining Compute Jobs are stopped
asynchronously. Compute Job failures can be tracked in the Ignite log for the
message `Remote job throws user exception…`.
```
##########
docs/_docs/distributed-computing/map-reduce.adoc:
##########
@@ -89,7 +89,17 @@ include::{javaFile}[tags=failover,indent=0]
== Compute Task Adapters
There are several helper classes that provide most commonly used
implementations of the `result(...)` and `map(...)` methods.
-* `ComputeTaskAdapter` — This class implements the `result()` method to return
the `FAILOVER` policy if a job throws an exception and the `WAIT` policy
otherwise. It means that this implementation will wait for all jobs to finish
with a result.
+* `ComputeTaskAdapter` — This class provides implementation for the `result()`
method, which makes the best effort to wait for all Compute Jobs to complete
successfully and process their results (see `WAIT` policy).
++
+--
+Comute Jobs will be automatically redeployed (see `FAILOVER` policy) in the
following scenarious:
Review Comment:
```suggestion
Compute Jobs will be automatically redeployed (see `FAILOVER` policy) in the
following scenarios:
```
##########
docs/_docs/distributed-computing/map-reduce.adoc:
##########
@@ -89,7 +89,17 @@ include::{javaFile}[tags=failover,indent=0]
== Compute Task Adapters
There are several helper classes that provide most commonly used
implementations of the `result(...)` and `map(...)` methods.
-* `ComputeTaskAdapter` — This class implements the `result()` method to return
the `FAILOVER` policy if a job throws an exception and the `WAIT` policy
otherwise. It means that this implementation will wait for all jobs to finish
with a result.
+* `ComputeTaskAdapter` — This class provides implementation for the `result()`
method, which makes the best effort to wait for all Compute Jobs to complete
successfully and process their results (see `WAIT` policy).
++
+--
+Comute Jobs will be automatically redeployed (see `FAILOVER` policy) in the
following scenarious:
+
+- the user-provided executor refused to process the mapped Compute Job;
+- the node intended to run the Compute Job is no longer present in the cluster;
+- the user code of the Compute Job explicitly threw a dedicated exception -
`ComputeJobFailoverException`.
+--
++
+If any of the mapped Compute Jobs threw a user exception, the entire Compute
Task immediately fails and all remaining Compute Jobs are stopped
asynchronously.
Review Comment:
```suggestion
If any of the mapped Compute Jobs throws a user exception, the entire
Compute Task immediately fails and all remaining Compute Jobs are stopped
asynchronously.
```
--
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.
To unsubscribe, e-mail: notifications-unsubscr...@ignite.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
