taklwu commented on a change in pull request #2237: URL: https://github.com/apache/hbase/pull/2237#discussion_r472482587
########## File path: hbase-server/src/main/java/org/apache/hadoop/hbase/master/procedure/InitMetaProcedure.java ########## @@ -167,4 +171,35 @@ protected void completionCleanup(MasterProcedureEnv env) { public void await() throws InterruptedException { latch.await(); } + + private static boolean deleteMetaTableDirectoryIfPartial(FileSystem rootDirectoryFs, + Path metaTableDir) throws IOException { + boolean isPartial = true; + try { + TableDescriptor metaDescriptor = + FSTableDescriptors.getTableDescriptorFromFs(rootDirectoryFs, metaTableDir); + // when entering the state of INIT_META_WRITE_FS_LAYOUT, if a meta table directory is found, + // the meta table should not have any useful data and considers as partial. + // if we find any valid HFiles, operator should fix the meta e.g. via HBCK. + if (metaDescriptor != null && metaDescriptor.getColumnFamilyCount() > 0) { + RemoteIterator<LocatedFileStatus> iterator = rootDirectoryFs.listFiles(metaTableDir, true); + while (iterator.hasNext()) { + LocatedFileStatus status = iterator.next(); + if (StoreFileInfo.isHFile(status.getPath()) && HFile + .isHFileFormat(rootDirectoryFs, status.getPath())) { + isPartial = false; + break; + } + } + } + } finally { + if (!isPartial) { + throw new IOException("Meta table is not partial, please sideline this meta directory " + + "or run HBCK to fix this meta table, e.g. rebuild the server hostname in ZNode for the " + + "meta region"); + } Review comment: @Apache9, we're failing the `InitMetaProcedure` with an `IOException`, and HMaster will fail the master startup if `InitMetaProcedure` is `FAILED` with an exception. Still, alternatively, we could continue the bootstrap without throwing (but this is not good as you recommended) so, do you think this change align with your comments? ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org