You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@hudi.apache.org by vb...@apache.org on 2020/01/21 05:07:52 UTC
[incubator-hudi] branch asf-site updated: [HUDI-403] Adds
guidelines on deployment/upgrading
This is an automated email from the ASF dual-hosted git repository.
vbalaji pushed a commit to branch asf-site
in repository https://gitbox.apache.org/repos/asf/incubator-hudi.git
The following commit(s) were added to refs/heads/asf-site by this push:
new 39db1ae [HUDI-403] Adds guidelines on deployment/upgrading
39db1ae is described below
commit 39db1aedbb9cb4533c58f869d5a940fbc1a3e5d2
Author: vinothchandar <vi...@apache.org>
AuthorDate: Mon Jan 20 18:01:19 2020 -0800
[HUDI-403] Adds guidelines on deployment/upgrading
- Moved "Adminsitering" page to "Deployment"
- Still need to add information on deltastreamer modes/compaction
---
docs/_data/navigation.yml | 6 +-
docs/_docs/2_2_writing_data.md | 2 +-
...{2_6_admin_guide.cn.md => 2_6_deployment.cn.md} | 2 +-
.../{2_6_admin_guide.md => 2_6_deployment.md} | 173 ++++++++++++---------
4 files changed, 107 insertions(+), 76 deletions(-)
diff --git a/docs/_data/navigation.yml b/docs/_data/navigation.yml
index d2826dd..a3510f4 100644
--- a/docs/_data/navigation.yml
+++ b/docs/_data/navigation.yml
@@ -38,8 +38,8 @@ docs:
url: /docs/configurations.html
- title: "Performance"
url: /docs/performance.html
- - title: "Administering"
- url: /docs/admin_guide.html
+ - title: "Deployment"
+ url: /docs/deployment.html
- title: INFO
children:
- title: "Docs Versions"
@@ -86,7 +86,7 @@ cn_docs:
- title: "性能"
url: /cn/docs/performance.html
- title: "管理"
- url: /cn/docs/admin_guide.html
+ url: /cn/docs/deployment.html
- title: 其他信息
children:
- title: "文档版本"
diff --git a/docs/_docs/2_2_writing_data.md b/docs/_docs/2_2_writing_data.md
index c0f5184..832daa6 100644
--- a/docs/_docs/2_2_writing_data.md
+++ b/docs/_docs/2_2_writing_data.md
@@ -8,7 +8,7 @@ last_modified_at: 2019-12-30T15:59:57-04:00
---
In this section, we will cover ways to ingest new changes from external sources or even other Hudi datasets using the [DeltaStreamer](#deltastreamer) tool, as well as
-speeding up large Spark jobs via upserts using the [Hudi datasource](#datasource-writer). Such datasets can then be [queried](querying_data.html) using various query engines.
+speeding up large Spark jobs via upserts using the [Hudi datasource](#datasource-writer). Such datasets can then be [queried](/docs/querying_data.html) using various query engines.
## Write Operations
diff --git a/docs/_docs/2_6_admin_guide.cn.md b/docs/_docs/2_6_deployment.cn.md
similarity index 99%
rename from docs/_docs/2_6_admin_guide.cn.md
rename to docs/_docs/2_6_deployment.cn.md
index ce055f0..c555b54 100644
--- a/docs/_docs/2_6_admin_guide.cn.md
+++ b/docs/_docs/2_6_deployment.cn.md
@@ -1,7 +1,7 @@
---
title: 管理 Hudi Pipelines
keywords: hudi, administration, operation, devops
-permalink: /cn/docs/admin_guide.html
+permalink: /cn/docs/deployment.html
summary: This section offers an overview of tools available to operate an ecosystem of Hudi datasets
toc: true
last_modified_at: 2019-12-30T15:59:57-04:00
diff --git a/docs/_docs/2_6_admin_guide.md b/docs/_docs/2_6_deployment.md
similarity index 73%
rename from docs/_docs/2_6_admin_guide.md
rename to docs/_docs/2_6_deployment.md
index 6990f50..295f8e8 100644
--- a/docs/_docs/2_6_admin_guide.md
+++ b/docs/_docs/2_6_deployment.md
@@ -1,51 +1,87 @@
---
-title: Administering Hudi Pipelines
-keywords: hudi, administration, operation, devops
-permalink: /docs/admin_guide.html
-summary: This section offers an overview of tools available to operate an ecosystem of Hudi datasets
+title: Deployment Guide
+keywords: hudi, administration, operation, devops, deployment
+permalink: /docs/deployment.html
+summary: This section offers an overview of tools available to operate an ecosystem of Hudi
toc: true
last_modified_at: 2019-12-30T15:59:57-04:00
---
-Admins/ops can gain visibility into Hudi datasets/pipelines in the following ways
+This section provides all the help you need to deploy and operate Hudi tables at scale.
+Specifically, we will cover the following aspects.
- - [Administering via the Admin CLI](#admin-cli)
- - [Graphite metrics](#metrics)
- - [Spark UI of the Hudi Application](#spark-ui)
+ - [Deployment Model](#deploying) : How various Hudi components are deployed and managed.
+ - [Upgrading Versions](#upgrading) : Picking up new releases of Hudi, guidelines and general best-practices
+ - [Migrating to Hudi](#migrating) : How to migrate your existing tables to Apache Hudi.
+ - [Interacting via CLI](#cli) : Using the CLI to perform maintenance or deeper introspection
+ - [Monitoring](#monitoring) : Tracking metrics from your hudi tables using popular tools.
+ - [Troubleshooting](#troubleshooting) : Uncovering, triaging and resolving issues in production.
+
+## Deploying
-This section provides a glimpse into each of these, with some general guidance on [troubleshooting](#troubleshooting)
+All in all, Hudi deploys with no long running servers or additional infrastructure cost to your data lake. In fact, Hudi pioneered this model of building a transactional distributed storage layer
+using existing infrastructure and its heartening to see other systems adopting similar approaches as well. Hudi writing is done via Spark jobs (DeltaStreamer or custom Spark datasource jobs), deployed per standard Apache Spark [recommendations](https://spark.apache.org/docs/latest/cluster-overview.html).
+Querying Hudi tables happens via libraries installed into Apache Hive, Apache Spark or Presto and hence no additional infrastructure is necessary.
-## Admin CLI
-Once hudi has been built, the shell can be fired by via `cd hudi-cli && ./hudi-cli.sh`.
-A hudi dataset resides on DFS, in a location referred to as the **basePath** and we would need this location in order to connect to a Hudi dataset.
-Hudi library effectively manages this dataset internally, using .hoodie subfolder to track all metadata
+## Upgrading
+
+New Hudi releases are listed on the [releases page](/releases), with detailed notes which list all the changes, with highlights in each release.
+At the end of the day, Hudi is a storage system and with that comes a lot of responsibilities, which we take seriously.
+
+As general guidelines,
+
+ - We strive to keep all changes backwards compatible (i.e new code can read old data/timeline files) and we cannot we will provide upgrade/downgrade tools via the CLI
+ - We cannot always guarantee forward compatibility (i.e old code being able to read data/timeline files written by a greater version). This is generally the norm, since no new features can be built otherwise.
+ However any large such changes, will be turned off by default, for smooth transition to newer release. After a few releases and once enough users deem the feature stable in production, we will flip the defaults in a subsequent release.
+ - Always upgrade the query bundles (mr-bundle, presto-bundle, spark-bundle) first and then upgrade the writers (deltastreamer, spark jobs using datasource). This often provides the best experience and it's easy to fix
+ any issues by rolling forward/back the writer code (which typically you might have more control over)
+ - With large, feature rich releases we recommend migrating slowly, by first testing in staging environments and running your own tests. Upgrading Hudi is no different than upgrading any database system.
+
+Note that release notes can override this information with specific instructions, applicable on case-by-case basis.
+
+## Migrating
+
+Currently migrating to Hudi can be done using two approaches
+
+- **Convert newer partitions to Hudi** : This model is suitable for large event tables (e.g: click streams, ad impressions), which also typically receive writes for the last few days alone. You can convert the last
+ N partitions to Hudi and proceed writing as if it were a Hudi table to begin with. The Hudi query side code is able to correctly handle both hudi and non-hudi data partitions.
+- **Full conversion to Hudi** : This model is suitable if you are currently bulk/full loading the table few times a day (e.g database ingestion). The full conversion of Hudi is simply a one-time step (akin to 1 run of your existing job),
+ which moves all of the data into the Hudi format and provides the ability to incrementally update for future writes.
+
+For more details, refer to the detailed [migration guide](/docs/migration_guide.html). In the future, we will be supporting seamless zero-copy bootstrap of existing tables with all the upsert/incremental query capabilities fully supported.
+
+## CLI
+
+Once hudi has been built, the shell can be fired by via `cd hudi-cli && ./hudi-cli.sh`. A hudi table resides on DFS, in a location referred to as the `basePath` and
+we would need this location in order to connect to a Hudi table. Hudi library effectively manages this table internally, using `.hoodie` subfolder to track all metadata
To initialize a hudi table, use the following command.
```java
-18/09/06 15:56:52 INFO annotation.AutowiredAnnotationBeanPostProcessor: JSR-330 'javax.inject.Inject' annotation found and supported for autowiring
-============================================
-* *
-* _ _ _ _ *
-* | | | | | | (_) *
-* | |__| | __| | - *
-* | __ || | / _` | || *
-* | | | || || (_| | || *
-* |_| |_|\___/ \____/ || *
-* *
-============================================
-
-Welcome to Hoodie CLI. Please type help if you are looking for help.
+===================================================================
+* ___ ___ *
+* /\__\ ___ /\ \ ___ *
+* / / / /\__\ / \ \ /\ \ *
+* / /__/ / / / / /\ \ \ \ \ \ *
+* / \ \ ___ / / / / / \ \__\ / \__\ *
+* / /\ \ /\__\ / /__/ ___ / /__/ \ |__| / /\/__/ *
+* \/ \ \/ / / \ \ \ /\__\ \ \ \ / / / /\/ / / *
+* \ / / \ \ / / / \ \ / / / \ /__/ *
+* / / / \ \/ / / \ \/ / / \ \__\ *
+* / / / \ / / \ / / \/__/ *
+* \/__/ \/__/ \/__/ Apache Hudi CLI *
+* *
+===================================================================
+
hudi->create --path /user/hive/warehouse/table1 --tableName hoodie_table_1 --tableType COPY_ON_WRITE
.....
-18/09/06 15:57:15 INFO table.HoodieTableMetaClient: Finished Loading Table of type COPY_ON_WRITE from ...
```
To see the description of hudi table, use the command:
```java
-hoodie:hoodie_table_1->desc
+hudi:hoodie_table_1->desc
18/09/06 15:57:19 INFO timeline.HoodieActiveTimeline: Loaded instants []
_________________________________________________________
| Property | Value |
@@ -58,37 +94,34 @@ hoodie:hoodie_table_1->desc
| hoodie.archivelog.folder| |
```
-Following is a sample command to connect to a Hudi dataset contains uber trips.
+Following is a sample command to connect to a Hudi table contains uber trips.
```java
-hoodie:trips->connect --path /app/uber/trips
+hudi:trips->connect --path /app/uber/trips
-16/10/05 23:20:37 INFO model.HoodieTableMetadata: Attempting to load the commits under /app/uber/trips/.hoodie with suffix .commit
-16/10/05 23:20:37 INFO model.HoodieTableMetadata: Attempting to load the commits under /app/uber/trips/.hoodie with suffix .inflight
16/10/05 23:20:37 INFO model.HoodieTableMetadata: All commits :HoodieCommits{commitList=[20161002045850, 20161002052915, 20161002055918, 20161002065317, 20161002075932, 20161002082904, 20161002085949, 20161002092936, 20161002105903, 20161002112938, 20161002123005, 20161002133002, 20161002155940, 20161002165924, 20161002172907, 20161002175905, 20161002190016, 20161002192954, 20161002195925, 20161002205935, 20161002215928, 20161002222938, 20161002225915, 20161002232906, 20161003003028, 201 [...]
Metadata for table trips loaded
-hoodie:trips->
```
-Once connected to the dataset, a lot of other commands become available. The shell has contextual autocomplete help (press TAB) and below is a list of all commands, few of which are reviewed in this section
+Once connected to the table, a lot of other commands become available. The shell has contextual autocomplete help (press TAB) and below is a list of all commands, few of which are reviewed in this section
are reviewed
```java
-hoodie:trips->help
+hudi:trips->help
* ! - Allows execution of operating system (OS) commands
* // - Inline comment markers (start of line only)
* ; - Inline comment markers (start of line only)
-* addpartitionmeta - Add partition metadata to a dataset, if not present
+* addpartitionmeta - Add partition metadata to a table, if not present
* clear - Clears the console
* cls - Clears the console
* commit rollback - Rollback a commit
-* commits compare - Compare commits with another Hoodie dataset
+* commits compare - Compare commits with another Hoodie table
* commit showfiles - Show file level details of a commit
* commit showpartitions - Show partition level details of a commit
* commits refresh - Refresh the commits
* commits show - Show the commits
-* commits sync - Compare commits with another Hoodie dataset
-* connect - Connect to a hoodie dataset
+* commits sync - Compare commits with another Hoodie table
+* connect - Connect to a hoodie table
* date - Displays the local date and time
* exit - Exits the shell
* help - List all commands usage
@@ -102,27 +135,26 @@ hoodie:trips->help
* utils loadClass - Load a class
* version - Displays shell version
-hoodie:trips->
+hudi:trips->
```
### Inspecting Commits
-The task of upserting or inserting a batch of incoming records is known as a **commit** in Hudi. A commit provides basic atomicity guarantees such that only commited data is available for querying.
+The task of upserting or inserting a batch of incoming records is known as a **commit** in Hudi. A commit provides basic atomicity guarantees such that only committed data is available for querying.
Each commit has a monotonically increasing string/number called the **commit number**. Typically, this is the time at which we started the commit.
To view some basic information about the last 10 commits,
```java
-hoodie:trips->commits show --sortBy "Total Bytes Written" --desc true --limit 10
+hudi:trips->commits show --sortBy "Total Bytes Written" --desc true --limit 10
________________________________________________________________________________________________________________________________________________________________________
| CommitTime | Total Bytes Written| Total Files Added| Total Files Updated| Total Partitions Written| Total Records Written| Total Update Records Written| Total Errors|
|=======================================================================================================================================================================|
....
....
....
-hoodie:trips->
```
At the start of each write, Hudi also writes a .inflight commit to the .hoodie folder. You can use the timestamp there to estimate how long the commit has been inflight
@@ -140,7 +172,7 @@ To understand how the writes spread across specific partiions,
```java
-hoodie:trips->commit showpartitions --commit 20161005165855 --sortBy "Total Bytes Written" --desc true --limit 10
+hudi:trips->commit showpartitions --commit 20161005165855 --sortBy "Total Bytes Written" --desc true --limit 10
__________________________________________________________________________________________________________________________________________
| Partition Path| Total Files Added| Total Files Updated| Total Records Inserted| Total Records Updated| Total Bytes Written| Total Errors|
|=========================================================================================================================================|
@@ -152,7 +184,7 @@ If you need file level granularity , we can do the following
```java
-hoodie:trips->commit showfiles --commit 20161005165855 --sortBy "Partition Path"
+hudi:trips->commit showfiles --commit 20161005165855 --sortBy "Partition Path"
________________________________________________________________________________________________________________________________________________________
| Partition Path| File ID | Previous Commit| Total Records Updated| Total Records Written| Total Bytes Written| Total Errors|
|=======================================================================================================================================================|
@@ -163,11 +195,11 @@ hoodie:trips->commit showfiles --commit 20161005165855 --sortBy "Partition Path"
### FileSystem View
-Hudi views each partition as a collection of file-groups with each file-group containing a list of file-slices in commit
-order (See Concepts). The below commands allow users to view the file-slices for a data-set.
+Hudi views each partition as a collection of file-groups with each file-group containing a list of file-slices in commit order (See [concepts]()).
+The below commands allow users to view the file-slices for a data-set.
```java
- hoodie:stock_ticks_mor->show fsview all
+hudi:stock_ticks_mor->show fsview all
....
_______________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________
| Partition | FileId | Base-Instant | Data-File | Data-File Size| Num Delta Files| Total Delta File Size| Delta Files |
@@ -176,24 +208,23 @@ order (See Concepts). The below commands allow users to view the file-slices for
- hoodie:stock_ticks_mor->show fsview latest --partitionPath "2018/08/31"
+hudi:stock_ticks_mor->show fsview latest --partitionPath "2018/08/31"
......
___________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________ [...]
| Partition | FileId | Base-Instant | Data-File | Data-File Size| Num Delta Files| Total Delta Size| Delta Size - compaction scheduled| Delta Size - compaction unscheduled| Delta To Base Ratio - compaction scheduled| Delta To Base Ratio - compaction unscheduled| Delta Files - compaction scheduled | Delta Files - compaction unscheduled|
|========================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================== [...]
| 2018/08/31| 111415c3-f26d-4639-86c8-f9956f245ac3| 20181002180759| hdfs://namenode:8020/user/hive/warehouse/stock_ticks_mor/2018/08/31/111415c3-f26d-4639-86c8-f9956f245ac3_0_20181002180759.parquet| 432.5 KB | 1 | 20.8 KB | 20.8 KB | 0.0 B | 0.0 B | 0.0 B | [HoodieLogFile {hdfs://namenode:8020/user/hive/warehouse/stock_ticks_mor/2018/08/31/.111415c3-f26d-4639-86c8-f9956f245ac3_20181002180759.log.1}]| [] |
- hoodie:stock_ticks_mor->
```
### Statistics
-Since Hudi directly manages file sizes for DFS dataset, it might be good to get an overall picture
+Since Hudi directly manages file sizes for DFS table, it might be good to get an overall picture
```java
-hoodie:trips->stats filesizes --partitionPath 2016/09/01 --sortBy "95th" --desc true --limit 10
+hudi:trips->stats filesizes --partitionPath 2016/09/01 --sortBy "95th" --desc true --limit 10
________________________________________________________________________________________________
| CommitTime | Min | 10th | 50th | avg | 95th | Max | NumFiles| StdDev |
|===============================================================================================|
@@ -206,7 +237,7 @@ In case of Hudi write taking much longer, it might be good to see the write ampl
```java
-hoodie:trips->stats wa
+hudi:trips->stats wa
__________________________________________________________________________
| CommitTime | Total Upserted| Total Written| Write Amplifiation Factor|
|=========================================================================|
@@ -227,7 +258,7 @@ To get an idea of the lag between compaction and writer applications, use the be
pending compactions.
```java
-hoodie:trips->compactions show all
+hudi:trips->compactions show all
___________________________________________________________________
| Compaction Instant Time| State | Total FileIds to be Compacted|
|==================================================================|
@@ -238,7 +269,7 @@ hoodie:trips->compactions show all
To inspect a specific compaction plan, use
```java
-hoodie:trips->compaction show --instant <INSTANT_1>
+hudi:trips->compaction show --instant <INSTANT_1>
_________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________
| Partition Path| File Id | Base Instant | Data File Path | Total Delta Files| getMetrics |
|================================================================================================================================================================================================================================================
@@ -249,11 +280,11 @@ hoodie:trips->compaction show --instant <INSTANT_1>
To manually schedule or run a compaction, use the below command. This command uses spark launcher to perform compaction
operations.
-**NOTE:** Make sure no other application is scheduling compaction for this dataset concurrently
+**NOTE:** Make sure no other application is scheduling compaction for this table concurrently
{: .notice--info}
```java
-hoodie:trips->help compaction schedule
+hudi:trips->help compaction schedule
Keyword: compaction schedule
Description: Schedule Compaction
Keyword: sparkMemory
@@ -266,7 +297,7 @@ Description: Schedule Compaction
```
```java
-hoodie:trips->help compaction run
+hudi:trips->help compaction run
Keyword: compaction run
Description: Run Compaction for given instant time
Keyword: tableName
@@ -300,7 +331,7 @@ Description: Run Compaction for given instant time
Default if unspecified: '__NULL__'
Keyword: compactionInstant
- Help: Base path for the target hoodie dataset
+ Help: Base path for the target hoodie table
Mandatory: true
Default if specified: '__NULL__'
Default if unspecified: '__NULL__'
@@ -313,7 +344,7 @@ Description: Run Compaction for given instant time
Validating a compaction plan : Check if all the files necessary for compactions are present and are valid
```java
-hoodie:stock_ticks_mor->compaction validate --instant 20181005222611
+hudi:stock_ticks_mor->compaction validate --instant 20181005222611
...
COMPACTION PLAN VALID
@@ -325,7 +356,7 @@ hoodie:stock_ticks_mor->compaction validate --instant 20181005222611
-hoodie:stock_ticks_mor->compaction validate --instant 20181005222601
+hudi:stock_ticks_mor->compaction validate --instant 20181005222601
COMPACTION PLAN INVALID
@@ -343,10 +374,10 @@ operation. Any new log-files that happened on this file after the compaction got
so that are preserved. Hudi provides the following CLI to support it
-### UnScheduling Compaction
+### Unscheduling Compaction
```java
-hoodie:trips->compaction unscheduleFileId --fileId <FileUUID>
+hudi:trips->compaction unscheduleFileId --fileId <FileUUID>
....
No File renames needed to unschedule file from pending compaction. Operation successful.
```
@@ -354,7 +385,7 @@ No File renames needed to unschedule file from pending compaction. Operation suc
In other cases, an entire compaction plan needs to be reverted. This is supported by the following CLI
```java
-hoodie:trips->compaction unschedule --compactionInstant <compactionInstant>
+hudi:trips->compaction unschedule --compactionInstant <compactionInstant>
.....
No File renames needed to unschedule pending compaction. Operation successful.
```
@@ -368,16 +399,16 @@ command comes to the rescue, it will rearrange the file-slices so that there is
consistent with the compaction plan
```java
-hoodie:stock_ticks_mor->compaction repair --instant 20181005222611
+hudi:stock_ticks_mor->compaction repair --instant 20181005222611
......
Compaction successfully repaired
.....
```
-## Metrics {#metrics}
+## Monitoring
-Once the Hudi Client is configured with the right datasetname and environment for metrics, it produces the following graphite metrics, that aid in debugging hudi datasets
+Once the Hudi writer is configured with the right table and environment for metrics, it produces the following graphite metrics, that aid in debugging hudi tables
- **Commit Duration** - This is amount of time it took to successfully commit a batch of records
- **Rollback Duration** - Similarly, amount of time taken to undo partial data left over by a failed commit (happens everytime automatically after a failing write)
@@ -392,7 +423,7 @@ These metrics can then be plotted on a standard tool like grafana. Below is a sa
</figure>
-## Troubleshooting Failures {#troubleshooting}
+## Troubleshooting
Section below generally aids in debugging Hudi failures. Off the bat, the following metadata is added to every record to help triage issues easily using standard Hadoop SQL engines (Hive/Presto/Spark)
@@ -400,9 +431,9 @@ Section below generally aids in debugging Hudi failures. Off the bat, the follow
- **_hoodie_commit_time** - Last commit that touched this record
- **_hoodie_file_name** - Actual file name containing the record (super useful to triage duplicates)
- **_hoodie_partition_path** - Path from basePath that identifies the partition containing this record
+
+ For performance related issues, please refer to the [tuning guide](https://cwiki.apache.org/confluence/display/HUDI/Tuning+Guide)
-**NOTE:** As of now, Hudi assumes the application passes in the same deterministic partitionpath for a given recordKey. i.e the uniqueness of record key is only enforced within each partition.
-{: .notice--info}
### Missing records
@@ -411,7 +442,7 @@ If you do find errors, then the record was not actually written by Hudi, but han
### Duplicates
-First of all, please confirm if you do indeed have duplicates **AFTER** ensuring the query is accessing the Hudi datasets [properly](sql_queries.html) .
+First of all, please confirm if you do indeed have duplicates **AFTER** ensuring the query is accessing the Hudi table [properly](/docs/sql_queries.html) .
- If confirmed, please use the metadata fields above, to identify the physical files & partition files containing the records .
- If duplicates span files across partitionpath, then this means your application is generating different partitionPaths for same recordKey, Please fix your app