[
https://issues.apache.org/jira/browse/HBASE-7987?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13593658#comment-13593658
]
Ted Yu commented on HBASE-7987:
-------------------------------
I think the idea of manifest is applicable to online snapshot as well.
See the following code snippet from HRegion.addRegionToSnapshot():
{code}
for (Store store : stores.values()) {
// 2.1. build the snapshot reference directory for the store
Path dstStoreDir = TakeSnapshotUtils.getStoreSnapshotDirectory(
snapshotRegionFs.getRegionDir(),
Bytes.toString(store.getFamily().getName()));
List<StoreFile> storeFiles = new
ArrayList<StoreFile>(store.getStorefiles());
if (LOG.isDebugEnabled()) {
LOG.debug("Adding snapshot references for " + storeFiles + " hfiles");
}
// 2.2. iterate through all the store's files and create "references".
int sz = storeFiles.size();
for (int i = 0; i < sz; i++) {
if (exnSnare != null) {
exnSnare.rethrowException();
}
Path file = storeFiles.get(i).getPath();
// create "reference" to this store file. It is intentionally an empty
file -- all
// necessary infomration is captured by its fs location and filename.
This allows us to
// only figure out what needs to be done via a single nn operation
(instead of having to
// open and read the files as well).
LOG.debug("Creating reference for file (" + (i+1) + "/" + sz + ") : " +
file);
Path referenceFile = new Path(dstStoreDir, file.getName());
boolean success = fs.getFileSystem().createNewFile(referenceFile);
{code}
Looking at the last line, we create reference file for each store file.
I think manifest file should be used in the above case as well.
> Snapshot Manifest file instead of multiple empty files
> ------------------------------------------------------
>
> Key: HBASE-7987
> URL: https://issues.apache.org/jira/browse/HBASE-7987
> Project: HBase
> Issue Type: Improvement
> Components: snapshots
> Reporter: Matteo Bertozzi
>
> Currently taking a snapshot means creating one empty file for each file in
> the source table directory, plus copying the .regioninfo file for each
> region, the table descriptor file and a snapshotInfo file.
> during the restore or snapshot verification we traverse the filesystem
> (fs.listStatus()) to find the snapshot files, and we open the .regioninfo
> files to get the information.
> to avoid hammering the NameNode and having lots of empty files, we can use a
> manifest file that contains the list of files and information that we need.
> To keep the RS parallelism that we have, each RS can write its own manifest.
> {code}
> message SnapshotDescriptor {
> required string name;
> optional string table;
> optional int64 creationTime;
> optional Type type;
> optional int32 version;
> }
> message SnapshotRegionManifest {
> optional int32 version;
> required RegionInfo regionInfo;
> repeated FamilyFiles familyFiles;
> message StoreFile {
> required string name;
> optional Reference reference;
> }
> message FamilyFiles {
> required bytes familyName;
> repeated StoreFile storeFiles;
> }
> }
> {code}
> {code}
> /hbase/.snapshot/<snapshotName>
> /hbase/.snapshot/<snapshotName>/snapshotInfo
> /hbase/.snapshot/<snapshotName>/<tableName>
> /hbase/.snapshot/<snapshotName>/<tableName>/tableInfo
> /hbase/.snapshot/<snapshotName>/<tableName>/regionManifest(.n)
> {code}
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
