[ https://issues.apache.org/jira/browse/HBASE-5878?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14270052#comment-14270052 ]
stack commented on HBASE-5878: ------------------------------ Is getFileLength same as getVisibleLength? What is minimum hadoop that has this API? Thanks. How about just letting the exception out: 120 SequenceFileLogReader.LOG.warn("Error while trying to get accurate file length. " 121 + "Truncation / data loss may occur if RegionServers die.", e); Since it is dataloss (as suggested above by [~umamaheswararao]?) Otherwise patch looks good. > Use getVisibleLength public api from HdfsDataInputStream from Hadoop-2. > ----------------------------------------------------------------------- > > Key: HBASE-5878 > URL: https://issues.apache.org/jira/browse/HBASE-5878 > Project: HBase > Issue Type: Bug > Components: wal > Reporter: Uma Maheswara Rao G > Assignee: Ashish Singhi > Fix For: 1.0.0, 2.0.0 > > Attachments: HBASE-5878.patch > > > SequencFileLogReader: > Currently Hbase using getFileLength api from DFSInputStream class by > reflection. DFSInputStream is not exposed as public. So, this may change in > future. Now HDFS exposed HdfsDataInputStream as public API. > We can make use of it, when we are not able to find the getFileLength api > from DFSInputStream as a else condition. So, that we will not have any sudden > surprise like we are facing today. > Also, it is just logging one warn message and proceeding if it throws any > exception while getting the length. I think we can re-throw the exception > because there is no point in continuing with dataloss. > {code} > long adjust = 0; > try { > Field fIn = FilterInputStream.class.getDeclaredField("in"); > fIn.setAccessible(true); > Object realIn = fIn.get(this.in); > // In hadoop 0.22, DFSInputStream is a standalone class. Before > this, > // it was an inner class of DFSClient. > if (realIn.getClass().getName().endsWith("DFSInputStream")) { > Method getFileLength = realIn.getClass(). > getDeclaredMethod("getFileLength", new Class<?> []{}); > getFileLength.setAccessible(true); > long realLength = ((Long)getFileLength. > invoke(realIn, new Object []{})).longValue(); > assert(realLength >= this.length); > adjust = realLength - this.length; > } else { > LOG.info("Input stream class: " + realIn.getClass().getName() + > ", not adjusting length"); > } > } catch(Exception e) { > SequenceFileLogReader.LOG.warn( > "Error while trying to get accurate file length. " + > "Truncation / data loss may occur if RegionServers die.", e); > } > return adjust + super.getPos(); > {code} -- This message was sent by Atlassian JIRA (v6.3.4#6332)