PollResultType
ClassAdLogReader::Poll() {
ProbeResultType probe_st;
FileOpErrCode fst;
fst = parser.openFile();
if(fst == FILE_OPEN_ERROR) {
#ifdef _NO_CONDOR_
syslog(LOG_ERR,
"Failed to open %s: errno=%d (%m)",
parser.getJobQueueName(), errno);
#else
dprintf(D_ALWAYS,"Failed to open %s: errno=%d\n",parser.getJobQueueName(),errno);
#endif
return POLL_FAIL;
}
probe_st = prober.probe(parser.getLastCALogEntry(),parser.getFilePointer());
bool success = true;
switch(probe_st) {
case COMPRESSED:
case PROBE_ERROR:
success = BulkLoad();
break;
case ADDITION:
success = IncrementalLoad();
break;
case NO_CHANGE:
break;
case PROBE_FATAL_ERROR:
return POLL_ERROR;
}
parser.closeFile();
if(success) {
// update prober to most recent observations about the job log file
prober.incrementProbeInfo();
}
return POLL_SUCCESS;
}
// load this M-tree with n data using the BulkLoad algorithm [CP98]
// data is an array of n entries
// padFactor is the maximum node utilization (use 1)
// name is the name of the tree
void
MT::BulkLoad (MTentry **data, int n, double padFactor, const char *name)
{
int size = 0;
if (EntrySize()) {
size = n * (sizeof(GiSTpage) + EntrySize()); // (only valid if we've fixed size entries)
} else {
for (int i=0; i<n; i++) {
size += sizeof(GiSTlte) + sizeof(GiSTpage) + data[i]->CompressedLength();
}
}
int totSize = size + GIST_PAGE_HEADER_SIZE + sizeof(GiSTlte);
if (totSize > Store()->PageSize()) { // we need to split the entries into several sub-trees
int numEntries = (int)(Store()->PageSize()*padFactor*n) / totSize;
int s = (int) MAX (MIN (numEntries, ceil(((float)n)/numEntries)), numEntries*MIN_UTIL); // initial number of samples
int nSamples, *samples = new int[s], *sizes = NULL, *ns = NULL, iter = 0, MAXITER = s * s;
GiSTlist<double *> *distm = (GiSTlist<double *> *) calloc (s, sizeof(GiSTlist<double *>)); // relative distances between samples
int MINSIZE = (int) (Store()->PageSize()*MIN_UTIL), addEntrySize = EntrySize() ? sizeof(GiSTpage) : sizeof(GiSTlte)+sizeof(GiSTpage);
GiSTlist<int> *lists = NULL; // set for each sample set
GiSTlist<double> *dists = NULL; // set for distance between each sample and its members
BOOL *bSampled = new BOOL[n]; // is this entry in the samples set?
// sampling phase
do {
iter++;
if (iter > 1) { // this is a new sampling phase
while (!lists[0].IsEmpty()) {
lists[0].RemoveFront ();
dists[0].RemoveFront ();
}
delete []lists;
delete []dists;
delete []sizes;
delete []ns;
while (!distm[0].IsEmpty()) {
delete []distm[0].RemoveFront(); // empty the distance list
}
for (int i=1; i<s; i++) {
distm[i].front = distm[i].rear = NULL;
}
}
if (iter >= MAXITER) {
cout << "Too many loops in BulkLoad!"<<endl<<"Please select a lower minimum node utilization or a bigger node size."<<endl;
exit(1);
}
for (int i=0; i<n; i++) {
bSampled[i] = FALSE;
}
nSamples = 0;
// pick s samples to create parents
while (nSamples < s) {
int i;
do {
i = PickRandom (0, n);
} while (bSampled[i]);
bSampled[i] = TRUE;
samples[nSamples++] = i;
}
lists = new GiSTlist<int>[s];
dists = new GiSTlist<double>[s];
sizes = new int[s];
ns = new int[s];
for (int i=0; i<s; i++) {
sizes[i] = GIST_PAGE_HEADER_SIZE + sizeof(GiSTlte);
ns[i] = 1;
distm[i].Prepend (new double[s]);
}
// compute the relative distances between samples
for (int i=0; i<s; i++) {
for (int j=0; j<i; j++) {
distm[j].front->entry[i] = distm[i].front->entry[j] = data[samples[j]]->object().distance(data[samples[i]]->object());
}
distm[i].front->entry[i] = 0;
}
// assign each entry to its nearest parent
for (int i=0; i<n; i++) {
if (bSampled[i]) {
int j = 0;
for (; samples[j]!=i; j++); // find this entry in the samples set and return position in it
lists[j].Prepend (i); // insert the entry in the right sample
dists[j].Prepend (0); // distance between sample and data[i]
sizes[j] += addEntrySize + data[i]->CompressedLength();
} else { // here we optimize the distance computations (like we do in the insert algorithm)
double *dist = new double[s]; // distance between this non-sample and samples
dist[0] = data[samples[0]]->object().distance(data[i]->object());
int minIndex = 0;
for (int j=1; j<s; j++) { // seek the nearest sample
dist[j] = -MaxDist();
if (fabs (data[samples[j]]->Key()->distance - data[i]->Key()->distance) >= dist[minIndex]) { // pruning
continue;
}
BOOL flag = TRUE;
for (int k=0; k<j && flag; k++) { // pruning (other samples)
if (dist[k] < 0) {
continue;
} else {
flag = fabs (dist[k] - distm[j].front->entry[k]) < dist[minIndex];
}
}
if (!flag) {
continue;
}
dist[j] = data[samples[j]]->object().distance(data[i]->object()); // have to compute this distance
if (dist[j] < dist[minIndex]) {
minIndex = j;
}
}
lists[minIndex].Append (i); // insert the entry in the right sample
dists[minIndex].Append (dist[minIndex]); // distance between sample and data[i]
sizes[minIndex] += addEntrySize + data[i]->CompressedLength();
ns[minIndex]++;
sizes[minIndex] >= MINSIZE ? delete []dist : distm[minIndex].Append (dist); // correspond with lists
}
}
// redistribute underfilled parents
int i;
while (sizes[i = FindMin (sizes, nSamples)] < MINSIZE) {
GiSTlist<int> list = lists[i]; // each sample set
while (!dists[i].IsEmpty()) { // clear distance between each sample and its members
dists[i].RemoveFront ();
}
// substitute this set with last set
for (int j=0; j<nSamples; j++) {
for (GiSTlistnode<double *> *node=distm[j].front; node; node=node->next) {
node->entry[i] = node->entry[nSamples-1];
}
}
GiSTlist<double *> dlist = distm[i]; // relative distances between sample[i] and other samples, reposition by myself
distm[i] = distm[nSamples-1];
lists[i] = lists[nSamples-1];
dists[i] = dists[nSamples-1];
samples[i] = samples[nSamples-1];
sizes[i] = sizes[nSamples-1];
ns[i] = ns[nSamples-1];
nSamples--;
while (!list.IsEmpty()) { // assign each entry to its nearest parent
double *dist = dlist.RemoveFront (); // relative distances between sample[i] (old) and other samples (old)
int minIndex = -1;
for (int j=0; j<nSamples && minIndex<0; j++) { // search for a computed distance
if (dist[j] > 0) {
minIndex = j;
}
}
int k = list.RemoveFront ();
if (minIndex < 0) { // no distance was computed (i.e. all distances were pruned)
dist[0] = data[samples[0]]->object().distance(data[k]->object());
minIndex = 0;
}
for (int j=0; j<nSamples; j++) {
if (j == minIndex) {
continue;
}
if (dist[j] < 0) { // distance wasn't computed
if (fabs (data[samples[j]]->Key()->distance - data[k]->Key()->distance) >= dist[minIndex]) {
continue; // pruning
}
BOOL flag = TRUE;
for (int i=0; i<j && flag; i++) { // pruning (other samples)
if (dist[i] < 0) {
continue;
} else {
flag = fabs (dist[i] - distm[j].front->entry[i]) < dist[minIndex];
}
}
if (!flag) {
continue;
}
dist[j] = data[samples[j]]->object().distance(data[k]->object()); // have to compute this distance
}
if (dist[j] < dist[minIndex]) {
minIndex = j;
}
}
lists[minIndex].Append (k);
dists[minIndex].Append (dist[minIndex]);
sizes[minIndex] += addEntrySize + data[k]->CompressedLength();
ns[minIndex]++;
sizes[minIndex] >= MINSIZE ? delete []dist : distm[minIndex].Append (dist); // correspond with lists
}
assert (dlist.IsEmpty()); // so is the list
}
} while (nSamples == 1); // if there's only one child, repeat the sampling phase
MTentry ***array = new MTentry **[nSamples]; // array of the entries for each sub-tree
for (int i=0; i<nSamples; i++) { // convert the lists into arrays
array[i] = new MTentry *[ns[i]];
for (int j=0; j<ns[i]; j++) {
array[i][j] = (MTentry *) data[lists[i].RemoveFront ()]->Copy();
array[i][j]->Key()->distance = dists[i].RemoveFront ();
}
assert (lists[i].IsEmpty());
assert (dists[i].IsEmpty());
}
delete []lists;
delete []dists;
delete []sizes;
delete []bSampled;
for (int i=0; i<nSamples; i++) {
while (!distm[i].IsEmpty()) {
delete [](distm[i].RemoveFront());
}
}
free (distm);
// build an M-tree under each parent
int nInit = nSamples;
MT *subtree = new MT;
GiSTlist<char *> subtreeNames; // list of the subtrees names
GiSTlist<MTentry *> topEntries; // list of the parent entries of each subtree
int nCreated = 0, minHeight = MAXINT;
char newName[50];
for (int i=0; i<nInit; i++) {
sprintf (newName, "%s.%i", name, ++nCreated);
unlink (newName);
subtree->Create(newName); // create the new subtree
subtree->BulkLoad(array[i], ns[i], padFactor, newName); // build the subtree
GiSTpath path;
path.MakeRoot ();
MTnode *subtreeRoot = (MTnode *) subtree->ReadNode(path);
if (subtreeRoot->IsUnderFull(*Store())) { // if the subtree root node is underfilled, we have to split the tree
GiSTlist<MTentry *> *parentEntries = new GiSTlist<MTentry *>;
GiSTlist<char *> *newTreeNames = subtree->SplitTree(&nCreated, subtree->TreeHeight()-1, parentEntries, name); // split the tree
nSamples--;
while (!newTreeNames->IsEmpty()) { // insert all the new trees in the subtrees list
subtreeNames.Append (newTreeNames->RemoveFront());
MTentry *entry = parentEntries->RemoveFront();
for (int j=0; j<n; j++) {
if (data[j]->object() == entry->object()) { // append the parent entry to the list
topEntries.Append (data[j]);
break;
}
}
delete entry;
nSamples++;
}
delete newTreeNames;
delete parentEntries;
minHeight = MIN (minHeight, subtree->TreeHeight()-1);
} else {
subtreeNames.Append (strdup(newName));
topEntries.Append (data[samples[i]]);
minHeight = MIN (minHeight, subtree->TreeHeight());
}
delete subtreeRoot;
subtree->Close();
delete subtree->Store(); // it was created in subtree->Create()
}
delete []samples;
for (int i=0; i<nInit; i++) {
for (int j=0; j<ns[i]; j++) {
delete array[i][j];
}
delete []array[i];
}
delete []array;
delete []ns;
// fix the subtree height
GiSTlist<char *> subtreeNames2; // list of the subtrees names
GiSTlist<MTentry *> topEntries2; // list of the parent entries of each subtree
while (!topEntries.IsEmpty()) { // insert the trees in the list (splitting trees if necessary)
MTentry *parentEntry = topEntries.RemoveFront ();
char *tmp = subtreeNames.RemoveFront ();
strcpy (newName, tmp);
delete []tmp;
subtree->Open(newName);
if (subtree->TreeHeight() > minHeight) { // we have to split the tree to reduce its height
nSamples--;
GiSTlist<MTentry *> *parentEntries = new GiSTlist<MTentry *>;
GiSTlist<char *> *newTreeNames = subtree->SplitTree(&nCreated, minHeight, parentEntries, name); // split the tree
while (!newTreeNames->IsEmpty()) { // insert all the new trees in the subtrees list
subtreeNames2.Append (newTreeNames->RemoveFront());
MTentry *entry = parentEntries->RemoveFront();
for (int j=0; j<n; j++) {
if (data[j]->object() == entry->object()) { // append the parent entry to the parents list
topEntries2.Append (data[j]);
break;;
}
}
delete entry;
nSamples++;
}
delete newTreeNames;
delete parentEntries;
} else { // simply insert the tree and its parent entry to the lists
subtreeNames2.Append (strdup(newName));
topEntries2.Append (parentEntry);
}
subtree->Close();
delete subtree->Store(); // it was created in tree->Open()
}
// build the super tree upon the parents
MTentry **topEntrArr = new MTentry *[nSamples]; // array of the parent entries for each subtree
char **subNameArr = new char *[nSamples]; // array of the subtrees names
for (int i=0; i<nSamples; i++) { // convert the lists into arrays
topEntrArr[i] = topEntries2.RemoveFront ();
subNameArr[i] = subtreeNames2.RemoveFront ();
}
assert (topEntries2.IsEmpty());
assert (subtreeNames2.IsEmpty());
sprintf (newName, "%s.0", name);
BulkLoad (topEntrArr, nSamples, padFactor, newName);
// attach each subtree to the leaves of the super tree
GiSTpath path;
path.MakeRoot ();
MTnode *node = (MTnode *) ReadNode (path);
GiSTlist<MTnode *> *oldList = new GiSTlist<MTnode *>; // upper level nodes
oldList->Append(node);
int level = node->Level();
while (level > 0) { // build the leaves list for super tree
GiSTlist<MTnode *> *newList = new GiSTlist<MTnode *>; // lower level nodes
while (!oldList->IsEmpty()) {
node = oldList->RemoveFront();
path = node->Path();
node->SetLevel(node->Level() + minHeight); // update level of the upper nodes of the super tree
WriteNode (node);
for (int i=0; i<node->NumEntries(); i++) {
MTentry *entry = (MTentry *) (*node)[i].Ptr();
path.MakeChild (entry->Ptr());
newList->Append((MTnode *)ReadNode(path));
path.MakeParent ();
}
delete node;
}
delete oldList;
oldList = newList;
level--;
}
while (!oldList->IsEmpty()) { // attach each subtree to its leaf
node = oldList->RemoveFront(); // retrieve next leaf (root of subtree)
node->SetLevel(minHeight); // update level of the root of the subtree
path = node->Path();
for (int i=0; i<node->NumEntries(); i++) {
MTentry *entry = (MTentry *) (*node)[i].Ptr();
path.MakeChild(Store()->Allocate());
MTnode *newNode = (MTnode *) CreateNode ();
newNode->Path() = path;
entry->SetPtr(path.Page());
path.MakeParent ();
int j = 0;
for (; entry->object() != topEntrArr[j]->object(); j++); // search the position to append
subtree->Open(subNameArr[j]);
GiSTpath rootPath;
rootPath.MakeRoot ();
Append (newNode, (MTnode *)subtree->ReadNode(rootPath)); // append this subtree to the super tree
subtree->Close();
delete subtree->Store(); // it was created in tree->Open()
delete newNode;
}
WriteNode (node);
delete node;
}
subtree->Open(subNameArr[0]); // in order to destroy the object tree
delete subtree;
for (int i=0; i<nSamples; i++) {
delete []subNameArr[i];
}
delete []subNameArr;
delete []topEntrArr;
// update radii of the upper nodes of the result M-tree
path.MakeRoot ();
node = (MTnode *) ReadNode (path);
oldList->Append(node);
level = node->Level();
while (level >= minHeight) { // build the list of the nodes which radii should be recomputed
GiSTlist<MTnode *> *newList = new GiSTlist<MTnode *>;
while (!oldList->IsEmpty()) {
node = oldList->RemoveFront();
path = node->Path();
for (int i=0; i<node->NumEntries(); i++) {
path.MakeChild ((*node)[i].Ptr()->Ptr());
newList->Append((MTnode *)ReadNode(path));
path.MakeParent ();
}
delete node;
}
delete oldList;
oldList = newList;
level--;
}
while (!oldList->IsEmpty()) { // adjust the radii of the nodes
MTnode *node = oldList->RemoveFront();
AdjKeys (node);
delete node;
}
delete oldList;
for (int i=0; i<=nCreated; i++) { // delete all temporary subtrees
sprintf (newName, "%s.%i", name, i);
unlink (newName);
}
} else { // we can insert all the entries in a single node
GiSTpath path;
path.MakeRoot ();
GiSTnode *node = ReadNode (path);
for (int i=0; i<n; i++) {
node->Insert(*(data[i]));
}
assert (!node->IsOverFull(*Store()));
WriteNode (node);
delete node;
}
}
