非常感谢张老师之前的解答，在ISSUE列表中我找到了如下patch: https://issues.apache.org/jira/browse/HBASE-26679 通过测试用例可以稳定复现我们使用版本的阻塞异常 https://issues.apache.org/jira/browse/HBASE-25905 这个是您提交的修复，wal sync阻塞时候会立即中断RegionServer进程，不知道我理解的是否有误，我先应用这两个修复，再观察下集群情况
leojie <leo...@apache.org> 于2023年5月15日周一 15:51写道： > 感谢张老师的回复，出现该问题期间，有张比较大的表在进行快照scan，资源并发数设置的比较大，引起read block > ops特别高，导致DN有压力（与历史对比，个别DN的XceiverCount指标持续异常高）， > 停止该任务后，这两天集群指标比较稳定了，没有发现WAL system stuck?的异常了。 > > 感觉DN有压力是引起此现象的一个导火索，但可能不是sync > wal操作hang住一两个小时后，才导致RS挂掉的直接原因，我试试升级jdk，再研读下这块的代码 > > 张铎(Duo Zhang) <palomino...@gmail.com> 于2023年5月13日周六 18:58写道： > >> 这看起来还是网络有抖动所以连接断了？建议还是查查出这个 log 时候集群的各种指标 >> >> 也可以尝试升级一下 JDK 吧 >> >> https://bugs.openjdk.org/browse/JDK-8215355 >> >> 这个是 8u251 之后才 fix 的，虽然这里写的是用 jemalloc 才会出现，但我之前也疑似遇到过，升级之后就再也没有了 >> >> leojie <leo...@apache.org> 于2023年5月11日周四 12:03写道： >> >> > java 的版本是：1.8.0_131 >> > >> > 翻了翻异常时间点之前的日志，发现有如下相关报错： >> > 2023-05-11 10:59:32,711 INFO [HBase-Metrics2-1] impl.MetricsSystemImpl: >> > Stopping HBase metrics system... >> > 2023-05-11 10:59:32,728 INFO [ganglia] impl.MetricsSinkAdapter: ganglia >> > thread interrupted. >> > 2023-05-11 10:59:32,728 INFO [HBase-Metrics2-1] impl.MetricsSystemImpl: >> > HBase metrics system stopped. >> > 2023-05-11 10:59:33,038 WARN >> > [AsyncFSWAL-0-hdfs://hadoop-bdxs-hb1-namenode/hbase] wal.AsyncFSWAL: >> sync >> > failed >> > java.io.IOException: stream already broken >> > at >> > org.apache.hadoop.hbase.io >> > >> .asyncfs.FanOutOneBlockAsyncDFSOutput.flush0(FanOutOneBlockAsyncDFSOutput.java:423) >> > at >> > org.apache.hadoop.hbase.io >> > >> .asyncfs.FanOutOneBlockAsyncDFSOutput.flush(FanOutOneBlockAsyncDFSOutput.java:512) >> > at >> > >> > >> org.apache.hadoop.hbase.regionserver.wal.AsyncProtobufLogWriter.sync(AsyncProtobufLogWriter.java:148) >> > at >> > >> > >> org.apache.hadoop.hbase.regionserver.wal.AsyncFSWAL.sync(AsyncFSWAL.java:379) >> > at >> > >> > >> org.apache.hadoop.hbase.regionserver.wal.AsyncFSWAL.consume(AsyncFSWAL.java:566) >> > at >> > >> > >> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142) >> > at >> > >> > >> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617) >> > at java.lang.Thread.run(Thread.java:748) >> > 2023-05-11 10:59:33,040 WARN >> > [AsyncFSWAL-0-hdfs://hadoop-bdxs-hb1-namenode/hbase] wal.AsyncFSWAL: >> sync >> > failed >> > java.io.IOException: stream already broken >> > at >> > org.apache.hadoop.hbase.io >> > >> .asyncfs.FanOutOneBlockAsyncDFSOutput.flush0(FanOutOneBlockAsyncDFSOutput.java:423) >> > at >> > org.apache.hadoop.hbase.io >> > >> .asyncfs.FanOutOneBlockAsyncDFSOutput.flush(FanOutOneBlockAsyncDFSOutput.java:512) >> > at >> > >> > >> org.apache.hadoop.hbase.regionserver.wal.AsyncProtobufLogWriter.sync(AsyncProtobufLogWriter.java:148) >> > at >> > >> > >> org.apache.hadoop.hbase.regionserver.wal.AsyncFSWAL.sync(AsyncFSWAL.java:379) >> > at >> > >> > >> org.apache.hadoop.hbase.regionserver.wal.AsyncFSWAL.consume(AsyncFSWAL.java:566) >> > at >> > >> > >> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142) >> > at >> > >> > >> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617) >> > at java.lang.Thread.run(Thread.java:748) >> > 2023-05-11 10:59:33,040 WARN >> > [AsyncFSWAL-0-hdfs://hadoop-bdxs-hb1-namenode/hbase] wal.AsyncFSWAL: >> sync >> > failed >> > >> org.apache.hbase.thirdparty.io.netty.channel.unix.Errors$NativeIoException: >> > syscall:read(..) failed: Connection reset by peer >> > at >> > >> > >> org.apache.hbase.thirdparty.io.netty.channel.unix.FileDescriptor.readAddress(..)(Unknown >> > Source) >> > 2023-05-11 10:59:33,040 WARN >> > [AsyncFSWAL-0-hdfs://hadoop-bdxs-hb1-namenode/hbase] wal.AsyncFSWAL: >> sync >> > failed >> > >> org.apache.hbase.thirdparty.io.netty.channel.unix.Errors$NativeIoException: >> > syscall:read(..) failed: Connection reset by peer >> > at >> > >> > >> org.apache.hbase.thirdparty.io.netty.channel.unix.FileDescriptor.readAddress(..)(Unknown >> > Source) >> > 2023-05-11 10:59:33,040 WARN >> > [AsyncFSWAL-0-hdfs://hadoop-bdxs-hb1-namenode/hbase] wal.AsyncFSWAL: >> sync >> > failed >> > >> org.apache.hbase.thirdparty.io.netty.channel.unix.Errors$NativeIoException: >> > syscall:read(..) failed: Connection reset by peer >> > at >> > >> > >> org.apache.hbase.thirdparty.io.netty.channel.unix.FileDescriptor.readAddress(..)(Unknown >> > Source) >> > 2023-05-11 10:59:33,040 WARN >> > [AsyncFSWAL-0-hdfs://hadoop-bdxs-hb1-namenode/hbase] wal.AsyncFSWAL: >> sync >> > failed >> > >> org.apache.hbase.thirdparty.io.netty.channel.unix.Errors$NativeIoException: >> > syscall:read(..) failed: Connection reset by peer >> > at >> > >> > >> org.apache.hbase.thirdparty.io.netty.channel.unix.FileDescriptor.readAddress(..)(Unknown >> > Source) >> > 2023-05-11 10:59:33,040 WARN >> > [AsyncFSWAL-0-hdfs://hadoop-bdxs-hb1-namenode/hbase] wal.AsyncFSWAL: >> sync >> > failed >> > >> org.apache.hbase.thirdparty.io.netty.channel.unix.Errors$NativeIoException: >> > syscall:read(..) failed: Connection reset by peer >> > at >> > >> > >> org.apache.hbase.thirdparty.io.netty.channel.unix.FileDescriptor.readAddress(..)(Unknown >> > Source) >> > 2023-05-11 10:59:33,040 WARN >> > [AsyncFSWAL-0-hdfs://hadoop-bdxs-hb1-namenode/hbase] wal.AsyncFSWAL: >> sync >> > failed >> > >> org.apache.hbase.thirdparty.io.netty.channel.unix.Errors$NativeIoException: >> > syscall:read(..) failed: Connection reset by peer >> > at >> > >> > >> org.apache.hbase.thirdparty.io.netty.channel.unix.FileDescriptor.readAddress(..)(Unknown >> > Source) >> > 2023-05-11 10:59:33,040 WARN >> > [AsyncFSWAL-0-hdfs://hadoop-bdxs-hb1-namenode/hbase] wal.AsyncFSWAL: >> sync >> > failed >> > >> org.apache.hbase.thirdparty.io.netty.channel.unix.Errors$NativeIoException: >> > syscall:read(..) failed: Connection reset by peer >> > at >> > >> > >> org.apache.hbase.thirdparty.io.netty.channel.unix.FileDescriptor.readAddress(..)(Unknown >> > Source) >> > 2023-05-11 10:59:33,040 WARN >> > [AsyncFSWAL-0-hdfs://hadoop-bdxs-hb1-namenode/hbase] wal.AsyncFSWAL: >> sync >> > failed >> > >> org.apache.hbase.thirdparty.io.netty.channel.unix.Errors$NativeIoException: >> > syscall:read(..) failed: Connection reset by peer >> > at >> > >> > >> org.apache.hbase.thirdparty.io.netty.channel.unix.FileDescriptor.readAddress(..)(Unknown >> > Source) >> > 2023-05-11 10:59:33,040 WARN >> > [AsyncFSWAL-0-hdfs://hadoop-bdxs-hb1-namenode/hbase] wal.AsyncFSWAL: >> sync >> > failed >> > >> org.apache.hbase.thirdparty.io.netty.channel.unix.Errors$NativeIoException: >> > syscall:read(..) failed: Connection reset by peer >> > at >> > >> > >> org.apache.hbase.thirdparty.io.netty.channel.unix.FileDescriptor.readAddress(..)(Unknown >> > Source) >> > 2023-05-11 10:59:33,040 WARN >> > [AsyncFSWAL-0-hdfs://hadoop-bdxs-hb1-namenode/hbase] wal.AsyncFSWAL: >> sync >> > failed >> > >> org.apache.hbase.thirdparty.io.netty.channel.unix.Errors$NativeIoException: >> > syscall:read(..) failed: Connection reset by peer >> > at >> > >> > >> org.apache.hbase.thirdparty.io.netty.channel.unix.FileDescriptor.readAddress(..)(Unknown >> > Source) >> > 2023-05-11 10:59:33,040 WARN >> > [AsyncFSWAL-0-hdfs://hadoop-bdxs-hb1-namenode/hbase] wal.AsyncFSWAL: >> sync >> > failed >> > >> org.apache.hbase.thirdparty.io.netty.channel.unix.Errors$NativeIoException: >> > syscall:read(..) failed: Connection reset by peer >> > at >> > >> > >> org.apache.hbase.thirdparty.io.netty.channel.unix.FileDescriptor.readAddress(..)(Unknown >> > Source) >> > 2023-05-11 10:59:33,040 WARN >> > [AsyncFSWAL-0-hdfs://hadoop-bdxs-hb1-namenode/hbase] wal.AsyncFSWAL: >> sync >> > failed >> > >> org.apache.hbase.thirdparty.io.netty.channel.unix.Errors$NativeIoException: >> > syscall:read(..) failed: Connection reset by peer >> > at >> > >> > >> org.apache.hbase.thirdparty.io.netty.channel.unix.FileDescriptor.readAddress(..)(Unknown >> > Source) >> > 2023-05-11 10:59:33,229 INFO [HBase-Metrics2-1] impl.MetricsConfig: >> Loaded >> > properties from hadoop-metrics2-hbase.properties >> > 2023-05-11 10:59:33,231 INFO [HBase-Metrics2-1] >> impl.MetricsSinkAdapter: >> > Sink ganglia started >> > 2023-05-11 10:59:33,257 INFO [HBase-Metrics2-1] impl.MetricsSystemImpl: >> > Scheduled Metric snapshot period at 20 second(s). >> > 2023-05-11 10:59:33,257 INFO [HBase-Metrics2-1] impl.MetricsSystemImpl: >> > HBase metrics system started >> > >> > >> > >> > 张铎(Duo Zhang) <palomino...@gmail.com> 于2023年5月11日周四 10:25写道： >> > >> > > 你往上翻翻有没有别的异常？这个看起来应该就是 AsyncFSWAL 有 bug 导致 hang 住不动了，不过我翻了一下，2.2.6 >> > > 之后似乎没有跟这个有关的 fix 了，之前倒是有一些。 >> > > >> > > 另外你的 jdk 版本是多少？我印象里 jdk8 早期版本 synchronized 有个 bug 可能会导致执行顺序错乱 >> > > >> > > leojie <leo...@apache.org> 于2023年5月9日周二 14:45写道： >> > > >> > > > hi all >> > > > 向社区求助一个问题，这两天总是在12:50左右遇到一个异常，描述如下： >> > > > hbase版本：2.2.6 >> > > > hadoop版本：3.3.1 >> > > > 异常现象：一个隔离组下的（只有一张表）的一个节点，在某一时刻write call >> > > > queue阻塞，阻塞时间点开始，这张表的读写qps都降为0，客户端读写不了该表，RS call >> > queue阻塞开始的时间点，日志中不断有如下报错： >> > > > 2023-05-08 12:42:27,310 ERROR [MemStoreFlusher.2] >> > > > regionserver.MemStoreFlusher: Cache flush failed for region >> > > > >> > > > >> > > >> > >> user_feature_v2,eacf_1658057555,1660314723816.2376cc2326b5372131cc530b115d959a. >> > > > org.apache.hadoop.hbase.exceptions.TimeoutIOException: Failed to get >> > sync >> > > > result after 300000 ms for txid=16920651960, WAL system stuck? >> > > > at >> > > > >> > > > >> > > >> > >> org.apache.hadoop.hbase.regionserver.wal.SyncFuture.get(SyncFuture.java:155) >> > > > at >> > > > >> > > > >> > > >> > >> org.apache.hadoop.hbase.regionserver.wal.AbstractFSWAL.blockOnSync(AbstractFSWAL.java:743) >> > > > at >> > > > >> > > > >> > > >> > >> org.apache.hadoop.hbase.regionserver.wal.AsyncFSWAL.sync(AsyncFSWAL.java:625) >> > > > at >> > > > >> > > > >> > > >> > >> org.apache.hadoop.hbase.regionserver.wal.AsyncFSWAL.sync(AsyncFSWAL.java:602) >> > > > at >> > > > >> > > > >> > > >> > >> org.apache.hadoop.hbase.regionserver.HRegion.doSyncOfUnflushedWALChanges(HRegion.java:2754) >> > > > at >> > > > >> > > > >> > > >> > >> org.apache.hadoop.hbase.regionserver.HRegion.internalPrepareFlushCache(HRegion.java:2691) >> > > > at >> > > > >> > > > >> > > >> > >> org.apache.hadoop.hbase.regionserver.HRegion.internalFlushcache(HRegion.java:2549) >> > > > at >> > > > >> > > > >> > > >> > >> org.apache.hadoop.hbase.regionserver.HRegion.internalFlushcache(HRegion.java:2523) >> > > > at >> > > > >> > > >> > >> org.apache.hadoop.hbase.regionserver.HRegion.flushcache(HRegion.java:2409) >> > > > at >> > > > >> > > > >> > > >> > >> org.apache.hadoop.hbase.regionserver.MemStoreFlusher.flushRegion(MemStoreFlusher.java:611) >> > > > at >> > > > >> > > > >> > > >> > >> org.apache.hadoop.hbase.regionserver.MemStoreFlusher.flushRegion(MemStoreFlusher.java:580) >> > > > at >> > > > >> > > > >> > > >> > >> org.apache.hadoop.hbase.regionserver.MemStoreFlusher.access$1000(MemStoreFlusher.java:68) >> > > > at >> > > > >> > > > >> > > >> > >> org.apache.hadoop.hbase.regionserver.MemStoreFlusher$FlushHandler.run(MemStoreFlusher.java:360) >> > > > at java.lang.Thread.run(Thread.java:748) >> > > > >> > > > >> > > >> > >> 节点memstore中无法刷新数据到WAL文件中，节点其他指标都正常，HDFS也没有压力。重启阻塞节点后，表恢复正常。异常期间，捕获的jstack文件我放进附件中了。 >> > > > 麻烦社区大佬有空帮忙定位下原因 >> > > > jstack 文件见ISSUE： https://issues.apache.org/jira/browse/HBASE-27850 >> 的附件 >> > > > >> > > >> > >> >
