removed re-read of index in case of a bad index. This may not solve the problem but it applies a 100% CPU problem on the peer. I'm afraid bad index files must be abandoned, and cannot be fixed this way.

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7111 6c8d7289-2bf4-0310-a012-ef5d649a1542
pull/1/head
orbiter 15 years ago
parent b2c9db48ea
commit d865ef77a8

@ -414,11 +414,12 @@ public class HeapReader {
file.readFully(keyf, 0, keyf.length);
if (!this.ordering.equal(key, keyf)) {
// verification of the indexed access failed. we must re-read the index
Log.logSevere("kelondroBLOBHeap", "verification indexed access for " + heapFile.toString() + " failed, re-building index");
Log.logSevere("kelondroBLOBHeap", "indexed verification access failed for " + heapFile.toString());
// this is a severe operation, it should never happen.
return null;
// but if the process ends in this state, it would completely fail
// if the index is not rebuild now at once
initIndexReadFromHeap();
//initIndexReadFromHeap();
}
// read the blob

Loading…
Cancel
Save