From 01876071b990063899e0839cdda9745783fe814c Mon Sep 17 00:00:00 2001 From: Michael Stack Date: Wed, 30 Sep 2020 09:34:15 -0700 Subject: [PATCH] HBASE-25109 Add MR Counters to WALPlayer; currently hard to tell if it is doing anything (#2468) Add MR counters so operator can see if WALPlayer run actually did anything. Fix bugs in usage (it enforced two args though usage describes allowing one arg only). Clean up usage output. In particular add note on wal file separator as hbase by default uses the ',' in its WAL file names which could befuddle operator trying to do simple import. Signed-off-by: Huaxiang Sun --- .../apache/hadoop/hbase/mapreduce/Driver.java | 3 +- .../hbase/mapreduce/WALInputFormat.java | 5 +- .../hadoop/hbase/mapreduce/WALPlayer.java | 72 ++++++++++++------- .../hadoop/hbase/mapreduce/TestWALPlayer.java | 4 +- src/main/asciidoc/_chapters/ops_mgt.adoc | 44 ++++++------ 5 files changed, 75 insertions(+), 53 deletions(-) diff --git a/hbase-mapreduce/src/main/java/org/apache/hadoop/hbase/mapreduce/Driver.java b/hbase-mapreduce/src/main/java/org/apache/hadoop/hbase/mapreduce/Driver.java index d52a31067f4..ed31c8422e7 100644 --- a/hbase-mapreduce/src/main/java/org/apache/hadoop/hbase/mapreduce/Driver.java +++ b/hbase-mapreduce/src/main/java/org/apache/hadoop/hbase/mapreduce/Driver.java @@ -1,4 +1,4 @@ -/** +/* * * Licensed to the Apache Software Foundation (ASF) under one * or more contributor license agreements. See the NOTICE file @@ -34,6 +34,7 @@ import org.apache.yetus.audience.InterfaceStability; @InterfaceAudience.LimitedPrivate(HBaseInterfaceAudience.TOOLS) @InterfaceStability.Stable public class Driver { + private Driver() {} public static void main(String[] args) throws Throwable { ProgramDriver pgd = new ProgramDriver(); diff --git a/hbase-mapreduce/src/main/java/org/apache/hadoop/hbase/mapreduce/WALInputFormat.java b/hbase-mapreduce/src/main/java/org/apache/hadoop/hbase/mapreduce/WALInputFormat.java index 1815412721f..7c4be83a73e 100644 --- a/hbase-mapreduce/src/main/java/org/apache/hadoop/hbase/mapreduce/WALInputFormat.java +++ b/hbase-mapreduce/src/main/java/org/apache/hadoop/hbase/mapreduce/WALInputFormat.java @@ -154,14 +154,13 @@ public class WALInputFormat extends InputFormat { WALSplit hsplit = (WALSplit)split; logFile = new Path(hsplit.getLogFileName()); conf = context.getConfiguration(); - LOG.info("Opening reader for "+split); + LOG.info("Opening {} for {}", logFile, split); openReader(logFile); this.startTime = hsplit.getStartTime(); this.endTime = hsplit.getEndTime(); } - private void openReader(Path path) throws IOException - { + private void openReader(Path path) throws IOException { closeReader(); reader = AbstractFSWALProvider.openReader(path, conf); seek(); diff --git a/hbase-mapreduce/src/main/java/org/apache/hadoop/hbase/mapreduce/WALPlayer.java b/hbase-mapreduce/src/main/java/org/apache/hadoop/hbase/mapreduce/WALPlayer.java index bbaa7549fa9..5b1aac65441 100644 --- a/hbase-mapreduce/src/main/java/org/apache/hadoop/hbase/mapreduce/WALPlayer.java +++ b/hbase-mapreduce/src/main/java/org/apache/hadoop/hbase/mapreduce/WALPlayer.java @@ -58,6 +58,8 @@ import org.apache.yetus.audience.InterfaceAudience; import org.slf4j.Logger; import org.slf4j.LoggerFactory; + + /** * A tool to replay WAL files as a M/R job. * The WAL can be replayed for a set of tables or all tables, @@ -140,7 +142,22 @@ public class WALPlayer extends Configured implements Tool { } /** - * A mapper that writes out {@link Mutation} to be directly applied to a running HBase instance. + * Enum for map metrics. Keep it out here rather than inside in the Map + * inner-class so we can find associated properties. + */ + protected static enum Counter { + /** Number of aggregated writes */ + PUTS, + /** Number of aggregated deletes */ + DELETES, + CELLS_READ, + CELLS_WRITTEN, + WALEDITS + } + + /** + * A mapper that writes out {@link Mutation} to be directly applied to + * a running HBase instance. */ protected static class WALMapper extends Mapper { @@ -148,6 +165,7 @@ public class WALPlayer extends Configured implements Tool { @Override public void map(WALKey key, WALEdit value, Context context) throws IOException { + context.getCounter(Counter.WALEDITS).increment(1); try { if (tables.isEmpty() || tables.containsKey(key.getTableName())) { TableName targetTable = @@ -157,6 +175,7 @@ public class WALPlayer extends Configured implements Tool { Delete del = null; Cell lastCell = null; for (Cell cell : value.getCells()) { + context.getCounter(Counter.CELLS_READ).increment(1); // Filtering WAL meta marker entries. if (WALEdit.isMetaEditFamily(cell)) { continue; @@ -172,9 +191,11 @@ public class WALPlayer extends Configured implements Tool { // row or type changed, write out aggregate KVs. if (put != null) { context.write(tableOut, put); + context.getCounter(Counter.PUTS).increment(1); } if (del != null) { context.write(tableOut, del); + context.getCounter(Counter.DELETES).increment(1); } if (CellUtil.isDelete(cell)) { del = new Delete(CellUtil.cloneRow(cell)); @@ -187,14 +208,17 @@ public class WALPlayer extends Configured implements Tool { } else { put.add(cell); } + context.getCounter(Counter.CELLS_WRITTEN).increment(1); } lastCell = cell; } // write residual KVs if (put != null) { context.write(tableOut, put); + context.getCounter(Counter.PUTS).increment(1); } if (del != null) { + context.getCounter(Counter.DELETES).increment(1); context.write(tableOut, del); } } @@ -270,7 +294,7 @@ public class WALPlayer extends Configured implements Tool { setupTime(conf, WALInputFormat.START_TIME_KEY); setupTime(conf, WALInputFormat.END_TIME_KEY); String inputDirs = args[0]; - String[] tables = args[1].split(","); + String[] tables = args.length == 1? new String [] {}: args[1].split(","); String[] tableMap; if (args.length > 2) { tableMap = args[2].split(","); @@ -278,7 +302,7 @@ public class WALPlayer extends Configured implements Tool { throw new IOException("The same number of tables and mapping must be provided."); } } else { - // if not mapping is specified map each table to itself + // if no mapping is specified, map each table to itself tableMap = tables; } conf.setStrings(TABLES_KEY, tables); @@ -349,27 +373,27 @@ public class WALPlayer extends Configured implements Tool { if (errorMsg != null && errorMsg.length() > 0) { System.err.println("ERROR: " + errorMsg); } - System.err.println("Usage: " + NAME + " [options] []"); - System.err.println("Replay all WAL files into HBase."); - System.err.println(" is a comma separated list of tables."); - System.err.println("If no tables (\"\") are specified, all tables are imported."); - System.err.println("(Be careful, hbase:meta entries will be imported in this case.)\n"); - System.err.println("WAL entries can be mapped to new set of tables via ."); - System.err.println(" is a comma separated list of target tables."); - System.err.println("If specified, each table in must have a mapping.\n"); - System.err.println("By default " + NAME + " will load data directly into HBase."); - System.err.println("To generate HFiles for a bulk data load instead, pass the following option:"); - System.err.println(" -D" + BULK_OUTPUT_CONF_KEY + "=/path/for/output"); - System.err.println(" (Only one table can be specified, and no mapping is allowed!)"); - System.err.println("Time range options:"); - System.err.println(" -D" + WALInputFormat.START_TIME_KEY + "=[date|ms]"); - System.err.println(" -D" + WALInputFormat.END_TIME_KEY + "=[date|ms]"); - System.err.println(" (The start and the end date of timerange. The dates can be expressed"); - System.err.println(" in milliseconds since epoch or in yyyy-MM-dd'T'HH:mm:ss.SS format."); - System.err.println(" E.g. 1234567890120 or 2009-02-13T23:32:30.12)"); + System.err.println("Usage: " + NAME + " [options] [ ]"); + System.err.println(" directory of WALs to replay."); + System.err.println(" comma separated list of tables. If no tables specified,"); + System.err.println(" all are imported (even hbase:meta if present)."); + System.err.println(" WAL entries can be mapped to a new set of tables by passing"); + System.err.println(" , a comma separated list of target tables."); + System.err.println(" If specified, each table in must have a mapping."); + System.err.println("To generate HFiles to bulk load instead of loading HBase directly, pass:"); + System.err.println(" -D" + BULK_OUTPUT_CONF_KEY + "=/path/for/output"); + System.err.println(" Only one table can be specified, and no mapping allowed!"); + System.err.println("To specify a time range, pass:"); + System.err.println(" -D" + WALInputFormat.START_TIME_KEY + "=[date|ms]"); + System.err.println(" -D" + WALInputFormat.END_TIME_KEY + "=[date|ms]"); + System.err.println(" The start and the end date of timerange. The dates can be expressed"); + System.err.println(" in milliseconds since epoch or in yyyy-MM-dd'T'HH:mm:ss.SS format."); + System.err.println(" E.g. 1234567890120 or 2009-02-13T23:32:30.12"); System.err.println("Other options:"); - System.err.println(" -D" + JOB_NAME_CONF_KEY + "=jobName"); - System.err.println(" Use the specified mapreduce job name for the wal player"); + System.err.println(" -D" + JOB_NAME_CONF_KEY + "=jobName"); + System.err.println(" Use the specified mapreduce job name for the wal player"); + System.err.println(" -Dwal.input.separator=' '"); + System.err.println(" Change WAL filename separator (WAL dir names use default ','.)"); System.err.println("For performance also consider the following options:\n" + " -Dmapreduce.map.speculative=false\n" + " -Dmapreduce.reduce.speculative=false"); @@ -387,7 +411,7 @@ public class WALPlayer extends Configured implements Tool { @Override public int run(String[] args) throws Exception { - if (args.length < 2) { + if (args.length < 1) { usage("Wrong number of arguments: " + args.length); System.exit(-1); } diff --git a/hbase-mapreduce/src/test/java/org/apache/hadoop/hbase/mapreduce/TestWALPlayer.java b/hbase-mapreduce/src/test/java/org/apache/hadoop/hbase/mapreduce/TestWALPlayer.java index 4880ab64e66..432aff1dd04 100644 --- a/hbase-mapreduce/src/test/java/org/apache/hadoop/hbase/mapreduce/TestWALPlayer.java +++ b/hbase-mapreduce/src/test/java/org/apache/hadoop/hbase/mapreduce/TestWALPlayer.java @@ -223,8 +223,8 @@ public class TestWALPlayer { } catch (SecurityException e) { assertEquals(-1, newSecurityManager.getExitCode()); assertTrue(data.toString().contains("ERROR: Wrong number of arguments:")); - assertTrue(data.toString().contains("Usage: WALPlayer [options] " + - " []")); + assertTrue(data.toString().contains("Usage: WALPlayer [options] " + + " [ ]")); assertTrue(data.toString().contains("-Dwal.bulk.output=/path/for/output")); } diff --git a/src/main/asciidoc/_chapters/ops_mgt.adoc b/src/main/asciidoc/_chapters/ops_mgt.adoc index d1bc2cf8e0c..6ea23655d3c 100644 --- a/src/main/asciidoc/_chapters/ops_mgt.adoc +++ b/src/main/asciidoc/_chapters/ops_mgt.adoc @@ -914,7 +914,7 @@ see <<_wal_tools>>. Invoke via: ---- -$ bin/hbase org.apache.hadoop.hbase.mapreduce.WALPlayer [options] []> +$ bin/hbase org.apache.hadoop.hbase.mapreduce.WALPlayer [options] [ ]> ---- For example: @@ -932,29 +932,27 @@ To NOT run WALPlayer as a mapreduce job on your cluster, force it to run all in Running `WALPlayer` with no arguments prints brief usage information: ---- -Usage: WALPlayer [options] [] -Replay all WAL files into HBase. - is a comma separated list of tables. -If no tables ("") are specified, all tables are imported. -(Be careful, hbase:meta entries will be imported in this case.) - -WAL entries can be mapped to new set of tables via . - is a comma separated list of target tables. -If specified, each table in must have a mapping. - -By default WALPlayer will load data directly into HBase. -To generate HFiles for a bulk data load instead, pass the following option: - -Dwal.bulk.output=/path/for/output - (Only one table can be specified, and no mapping is allowed!) -Time range options: - -Dwal.start.time=[date|ms] - -Dwal.end.time=[date|ms] - (The start and the end date of timerange. The dates can be expressed - in milliseconds since epoch or in yyyy-MM-dd'T'HH:mm:ss.SS format. - E.g. 1234567890120 or 2009-02-13T23:32:30.12) +Usage: WALPlayer [options] [ ] + directory of WALs to replay. + comma separated list of tables. If no tables specified, + all are imported (even hbase:meta if present). + WAL entries can be mapped to a new set of tables by passing + , a comma separated list of target tables. + If specified, each table in must have a mapping. +To generate HFiles to bulk load instead of loading HBase directly, pass: + -Dwal.bulk.output=/path/for/output + Only one table can be specified, and no mapping allowed! +To specify a time range, pass: + -Dwal.start.time=[date|ms] + -Dwal.end.time=[date|ms] + The start and the end date of timerange. The dates can be expressed + in milliseconds since epoch or in yyyy-MM-dd'T'HH:mm:ss.SS format. + E.g. 1234567890120 or 2009-02-13T23:32:30.12 Other options: - -Dmapreduce.job.name=jobName - Use the specified mapreduce job name for the wal player + -Dmapreduce.job.name=jobName + Use the specified mapreduce job name for the wal player + -Dwal.input.separator=' ' + Change WAL filename separator (WAL dir names use default ','.) For performance also consider the following options: -Dmapreduce.map.speculative=false -Dmapreduce.reduce.speculative=false