[
https://issues.apache.org/jira/browse/HBASE-16644?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Mikhail Antonov updated HBASE-16644:
------------------------------------
Attachment: HBASE-16644.branch-1.3.patch
Ok, the first draft version of the patch broke HFIle tool, bulkload and other
codepaths not going through Store* interfaces, since those don't use
verifyChecksum flag as expected. So I digged in and found that before that
major refactoring of this codebase we used to use
hfileContext.isHBaseChecksum..() call to verify whether we should be using
checksums or not (which simply checks minor HFile version from the trailer).
With updated patch, i'm able to handle both 2.0 and new files. The tests broken
by previous version of the patch pass for me now. pinging [~stack]
> Errors when reading legit HFile' Trailer on branch 1.3
> ------------------------------------------------------
>
> Key: HBASE-16644
> URL: https://issues.apache.org/jira/browse/HBASE-16644
> Project: HBase
> Issue Type: Bug
> Components: HFile
> Affects Versions: 1.3.0, 1.4.0
> Reporter: Mikhail Antonov
> Assignee: Mikhail Antonov
> Priority: Critical
> Fix For: 1.3.0
>
> Attachments: HBASE-16644.branch-1.3.patch,
> HBASE-16644.branch-1.3.patch
>
>
> There seems to be a regression in branch 1.3 where we can't read HFile
> trailer(getting "CorruptHFileException: Problem reading HFile Trailer") on
> some HFiles that could be successfully read on 1.2.
> I've seen this error manifesting in two ways so far.
> {code}Caused by: org.apache.hadoop.hbase.io.hfile.CorruptHFileException:
> Problem reading HFile Trailer from file <file name>
> at
> org.apache.hadoop.hbase.io.hfile.HFile.pickReaderVersion(HFile.java:497)
> at org.apache.hadoop.hbase.io.hfile.HFile.createReader(HFile.java:525)
> at
> org.apache.hadoop.hbase.regionserver.StoreFile$Reader.<init>(StoreFile.java:1164)
> at
> org.apache.hadoop.hbase.regionserver.StoreFileInfo.open(StoreFileInfo.java:259)
> at
> org.apache.hadoop.hbase.regionserver.StoreFile.open(StoreFile.java:427)
> at
> org.apache.hadoop.hbase.regionserver.StoreFile.createReader(StoreFile.java:528)
> at
> org.apache.hadoop.hbase.regionserver.StoreFile.createReader(StoreFile.java:518)
> at
> org.apache.hadoop.hbase.regionserver.HStore.createStoreFileAndReader(HStore.java:652)
> at
> org.apache.hadoop.hbase.regionserver.HStore.access$000(HStore.java:117)
> at org.apache.hadoop.hbase.regionserver.HStore$1.call(HStore.java:519)
> at org.apache.hadoop.hbase.regionserver.HStore$1.call(HStore.java:516)
> ... 6 more
> Caused by: java.io.IOException: Invalid HFile block magic:
> \x00\x00\x04\x00\x00\x00\x00\x00
> at org.apache.hadoop.hbase.io.hfile.BlockType.parse(BlockType.java:155)
> at org.apache.hadoop.hbase.io.hfile.BlockType.read(BlockType.java:167)
> at
> org.apache.hadoop.hbase.io.hfile.HFileBlock.<init>(HFileBlock.java:344)
> at
> org.apache.hadoop.hbase.io.hfile.HFileBlock$FSReaderImpl.readBlockDataInternal(HFileBlock.java:1735)
> at
> org.apache.hadoop.hbase.io.hfile.HFileBlock$FSReaderImpl.readBlockData(HFileBlock.java:1558)
> at
> org.apache.hadoop.hbase.io.hfile.HFileBlock$AbstractFSReader$1.nextBlock(HFileBlock.java:1397)
> at
> org.apache.hadoop.hbase.io.hfile.HFileBlock$AbstractFSReader$1.nextBlockWithBlockType(HFileBlock.java:1405)
> at
> org.apache.hadoop.hbase.io.hfile.HFileReaderV2.<init>(HFileReaderV2.java:156)
> at
> org.apache.hadoop.hbase.io.hfile.HFile.pickReaderVersion(HFile.java:485)
> {code}
> and second
> {code}
> Caused by: org.apache.hadoop.hbase.io.hfile.CorruptHFileException: Problem
> reading HFile Trailer from file <file path>
> at
> org.apache.hadoop.hbase.io.hfile.HFile.pickReaderVersion(HFile.java:497)
> at org.apache.hadoop.hbase.io.hfile.HFile.createReader(HFile.java:525)
> at
> org.apache.hadoop.hbase.regionserver.StoreFile$Reader.<init>(StoreFile.java:1164)
> at
> org.apache.hadoop.hbase.io.HalfStoreFileReader.<init>(HalfStoreFileReader.java:104)
> at
> org.apache.hadoop.hbase.regionserver.StoreFileInfo.open(StoreFileInfo.java:256)
> at
> org.apache.hadoop.hbase.regionserver.StoreFile.open(StoreFile.java:427)
> at
> org.apache.hadoop.hbase.regionserver.StoreFile.createReader(StoreFile.java:528)
> at
> org.apache.hadoop.hbase.regionserver.StoreFile.createReader(StoreFile.java:518)
> at
> org.apache.hadoop.hbase.regionserver.HStore.createStoreFileAndReader(HStore.java:652)
> at
> org.apache.hadoop.hbase.regionserver.HStore.access$000(HStore.java:117)
> at org.apache.hadoop.hbase.regionserver.HStore$1.call(HStore.java:519)
> at org.apache.hadoop.hbase.regionserver.HStore$1.call(HStore.java:516)
> ... 6 more
> Caused by: java.io.IOException: Premature EOF from inputStream (read returned
> -1, was trying to read 10083 necessary bytes and 24 extra bytes, successfully
> read 1072
> at
> org.apache.hadoop.hbase.io.hfile.HFileBlock.readWithExtra(HFileBlock.java:737)
> at
> org.apache.hadoop.hbase.io.hfile.HFileBlock$AbstractFSReader.readAtOffset(HFileBlock.java:1459)
> at
> org.apache.hadoop.hbase.io.hfile.HFileBlock$FSReaderImpl.readBlockDataInternal(HFileBlock.java:1712)
> at
> org.apache.hadoop.hbase.io.hfile.HFileBlock$FSReaderImpl.readBlockData(HFileBlock.java:1558)
> at
> org.apache.hadoop.hbase.io.hfile.HFileBlock$AbstractFSReader$1.nextBlock(HFileBlock.java:1397)
> at
> org.apache.hadoop.hbase.io.hfile.HFileBlock$AbstractFSReader$1.nextBlockWithBlockType(HFileBlock.java:1405)
> at
> org.apache.hadoop.hbase.io.hfile.HFileReaderV2.<init>(HFileReaderV2.java:156)
> at
> org.apache.hadoop.hbase.io.hfile.HFile.pickReaderVersion(HFile.java:485)
> {code}
> In my case this problem was reproducible by running `hbase hfile -m -f`
> command. There seem to be two changes in behavior introduced by HBASE-15477
> (cc [~stack]).
> One is that HFileBlock#getOnDiskSizeWithHeader always assumes that checksum
> verification is true (this behavior results in onDiskSizeWithHeader being
> calculated differently in 1.2 and 1.3 for some cases).
> Second is that in HFileBlock constructor we always attempt to retrive
> checksum-related fields from the header even if no checksum verification is
> going on.
> Attached is the patch which when applied allows 1.3 to read again the same
> HFiles.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
