Recovery Process

Recovery are performed at the granularity of the storage group, and the entry point for recovery is the recover() of the StorageGroupProcessor

Recovery Process Of Storage Group

  • First get all the data files ending with .tsfile in the storage group, return TsFileResource, there are several file lists as follows

    • Sequence Files
      • 0.10 version tsfiles（sealed/unsealed）
      • 0.9 version tsfiles（sealed）
    • Unsequence Files
      • 0.10 version tsfiles（sealed/unsealed）
      • 0.9 version tsfiles（sealed）

  • If there exists 0.9 version TsFile in the storage group, add the old version's sequence and unsequence files to upgradeSeqFileList and upgradeSeqFileList respectively for upgrade and query.

  • Group sequence and unsequence files according to partition -- Map<Long, List<TsFileResource>>

  • To recover the sequential files of each partition, take the sequential TsFile of each partition obtained in the previous step as a parameter, and call recoverTsFiles to recover. This method will put the restored sequence TsFile into the sequenceFileTreeSet in the form of TsFileResource. If the TsFile is the last one of this partition and it is not sealed, construct a TsFileProcessor object for it and add it to workSequenceTsFileProcessors. The specific details of this method will be explained in the next section.

  • To recover the disordered files of each partition, take the unsequential TsFile of each partition obtained in the previous step as a parameter, and call recoverTsFiles to recover. This method will put the restored unsequence TsFile into the unSequenceFileList in the form of TsFileResource. if the TsFile is the last one in this partition and it is not sealed, a TsFileProcessor object must be constructed for it and added to workUnsequenceTsFileProcessors. The specific details of this method will be explained in the next section.

  • Traverse the sequenceFileTreeSet and unSequenceFileList obtained in the previous two steps respectively, and update the version number corresponding to the partition

  • Check whether there is a Modification file during the merge, and call the RecoverMergeTask.recoverMerge method to recover the merge

  • Call the updateLastestFlushedTime() method to update the latestTimeForEachDevice, partitionLatestFlushedTimeForEachDevice and globalLatestFlushedTimeForEachDevice with sequential tsfile of version 0.9

    • latestTimeForEachDevice records the latest timestamp under each partition that all devices have been inserted into (including unflushed and flushed)
    • partitionLatestFlushedTimeForEachDevice records the latest timestamp of all devices under each partition that has been flushed. It is used to determine whether a newly inserted point is out of order.
    • globalLatestFlushedTimeForEachDevice records the latest timestamp of all devices that have been flushed (a summary of the latest timestamps of each partition)

  • Finally traverse sequenceFileTreeSet, and use the restored sequence file to update latestTimeForEachDevice, partitionLatestFlushedTimeForEachDevice and globalLatestFlushedTimeForEachDevice again

Recover a TsFile(Seq/Unseq) of each partiton

  • org.apache.iotdb.db.engine.storagegroup.StorageGroupProcessor.recoverTsFiles

This method is mainly responsible for traversing all TsFiles passed in and recovering them one by one.

  • Construct a TsFileRecoverPerformer object to recover the TsFile. The recovery logic is encapsulated in the recover() method of TsFileRecoverPerformer (details will be explained in the next section), which will return a restored RestorableTsFileIOWriter Object.

    • If the recovery process fails, record the log and skip the tsfile

  • If the TsFile file is not the last file, or the TsFile file is the last file, but has been closed or marked as closed, just set the closed attribute of the TsFileResource object corresponding to the TsFile in memory to true.

  • If the TsFile file can continue to be written, it means that this is the last TsFile of this partition, and it is unsealed, and it will continue to remain unsealed. You need to construct a TsFileProcessor object for it and place it in workSequenceTsFileProcessors or workUnsequenceTsFileProcessors.

  • Finally, put the corresponding TsFileResource object in the memory of the restored TsFile into sequenceFileTreeSet or unSequenceFileList

Details about recovering a TsFile

  • org.apache.iotdb.db.writelog.recover.TsFileRecoverPerformer.recover

This method is mainly responsible for the recovery of each specific TsFile file.

  • First use the tsfile to construct a RestorableTsFileIOWriter object. In the construction method of RestorableTsFileIOWriter, the content of the tsfile will be checked and truncated if necessary

    • If there is nothing in this file, write MAGIC_STRING and VERSION_NUMBER for it, and return directly. At this time, crashed is false, and canWrite is true;
    • If there is content in this file, construct a TsFileSequenceReader object to parse the content, call the selfCheck method, truncate the incomplete content and initialize truncatedSize to HeaderLength
      • If the content of the file is complete (have a complete header of MAGIC_STRING and VERSION_NUMBER, and a tail of MAGIC_STRING), return TsFileCheckStatus.COMPLETE_FILE
      • If the file length is less than HeaderLength(len(MAGIC_STRING) + len(VERSION_NUMBER)), or the content of the file header is not MAGIC_STRING + VERSION_NUMBER , return INCOMPATIBLE_FILE
      • If the file length is exactly equal to HeaderLength, and the file content is MAGIC_STRING + VERSION_NUMBER, then retunr HeaderLength
      • If the file length is greater than HeaderLength and the file header is legal, but there is no MAGIC_STRING at the end of the file, it means that the file is incomplete and needs to be truncated. Read from VERSION_NUMBER position, read out the data in the following chunk, and recover the ChunkMetadata based on the data in the chunk. If you encounter CHUNK_GROUP_FOOTER, it means that the entire ChunkGroup is complete. Update truncatedSize to the current position
      • Return truncatedSize
    • truncated the file according to the returned truncatedSize
      • If truncatedSize is equal to TsFileCheckStatus.COMPLETE_FILE, set crashed and canWrite to false, and close the output stream of the file
      • If truncatedSize is equal to TsFileCheckStatus.INCOMPATIBLE_FILE, the output stream of the file is closed and an exception is thrown
      • Otherwise, set crashed and canWrite to true and truncated the file to truncatedSize

  • Judge whether the file is complete by the returned RestorableTsFileIOWriter

    • If the TsFile file is complete

      • If the resource file corresponding to the TsFile exists, the resource file is deserialized (including the minimum and maximum timestamps of each device in the tsfile), and the file version number is restored
      • If the resource file corresponding to the TsFile does not exist, regenerate the resource file and persist it to disk.
      • Return the generated RestorableTsFileIOWriter

    • If TsFile is incomplete

      • Call recoverResourceFromWriter to recover the resource information through the ChunkMetadata information in RestorableTsFileIOWriter
      • Call the redoLogs method to write the data in one or more wal files corresponding to this file to a temporary Memtable and persist to this incomplete TsFile
        • For sequential files, skip WALs whose timestamp is less than or equal to the current resource
        • For unsequential files, redo all WAL, it is possible to repeatedly write to ChunkGroup of multiple devices
      • If the TsFile is not the last TsFile of the current partition, or there is a .closing file in the TsFile, call the endFile() method of RestorableTsFileIOWriter to seal the file, delete the .closing file and generates resource file for it.