Github user mfenes commented on a diff in the pull request:
https://github.com/apache/zookeeper/pull/450#discussion_r163282090
--- Diff:
src/java/main/org/apache/zookeeper/server/persistence/FileTxnSnapLog.java ---
@@ -136,13 +136,51 @@ public FileTxnSnapLog(File dataDir, File snapDir)
throws IOException {
throw new DatadirException("Cannot write to snap directory " +
this.snapDir);
}
+ // check content of transaction log and snapshot dirs if they are
two different directories
+ if(!this.dataDir.getPath().equals(this.snapDir.getPath())){
+ checkLogDir();
+ checkSnapDir();
+ }
+
txnLog = new FileTxnLog(this.dataDir);
snapLog = new FileSnap(this.snapDir);
autoCreateDB =
Boolean.parseBoolean(System.getProperty(ZOOKEEPER_DB_AUTOCREATE,
ZOOKEEPER_DB_AUTOCREATE_DEFAULT));
}
+ private void checkLogDir() throws LogdirContentCheckException {
+ File[] files = this.dataDir.listFiles();
--- End diff --
If I used FilenameFilter, then Util.isSnapshotFile() / Util.isLogFile()
check would be run for all the files in the directory and
listFiles(FilenameFilter filter) would return all the files satisfying the
filter condition, however I need only the first occurrence which satisfies the
condition, not all of them. The current logic quits from the for loop
immediately when it finds a file violating the configuration and throws an
exception, while your proposal would iterate over all the files in the
directory and would call Util.isSnapshotFile() / Util.isLogFile() for each of
the files inside FilenameFilter to prepare the filtered File[]. So using
FilenameFilter would be a bit slower, but yes, it might need less lines in
code, also at the price of obscuring the purpose of the code (i.e.
hasSnapshotFiles / hasLogFiles boolean variables tell what the problem exactly
is, while if (snapshotFiles.length > 0) { throw new Exception(...) } would
not). However, if we prefer using Java library
classes over standard coding patterns even in cases when it does not fit the
purpose entirely, then FilenameFilter can be the winner.
---
