/*
* Returns true iff
*/
bool
CacheStrategyUtility::isResponsibleForDataObject(
DataObjectRef &dObj)
{
string id = string(dObj->getIdStr());
/*
// THIS IS NOT THREAD SAFE!! added getOrigSize
// to grab unaltered file size.
if (utilMetadata.find(id) != utilMetadata.end()) {
return true;
}
*/
if (dObj->isDuplicate()) {
return false;
}
// SW: TODO: NOTE: this might not be the best way to check if it's from
// a local application, but it works for now...
bool isLocal = dObj->getRemoteInterface() && dObj->getRemoteInterface()->isApplication();
bool notResponsible = dObj->isControlMessage() || dObj->isNodeDescription();
bool isResponsible = !notResponsible;
if (!handle_zero_size) {
isResponsible = isResponsible && (dObj->getOrigDataLen() > 0);
}
if (!manage_locally_sent_files) {
isResponsible = isResponsible && dObj->isPersistent();
}
if (stats_replacement_strat && stats_replacement_strat->isResponsibleForDataObject(dObj)) {
isResponsible = true;
} else if (manage_only_remote_files) {
isResponsible = isResponsible && !isLocal;
}
return isResponsible;
}
/*
* Callback when a new data object is inserted. Updates the utility strategy
* state and notifies the utility functions.
*/
void
CacheStrategyUtility::_handleNewDataObject(
DataObjectRef &dObj)
{
if (!isResponsibleForDataObject(dObj)) {
//HAGGLE_DBG("Ignoring data object, in-eligible for caching\n");
return;
}
if (stats_replacement_strat && stats_replacement_strat->isResponsibleForDataObject(dObj)) {
stats_replacement_strat->handleNewDataObject(dObj);
return;
}
string id = string(dObj->getIdStr());
if (utilMetadata.find(id) != utilMetadata.end()) {
// the DO is already in the cache!
HAGGLE_DBG("Received data object already in the cache: %s\n", id.c_str());
current_dupe_do_recv++;
return;
}
getUtilityFunction()->notifyInsertion(dObj);
int cost = dObj->getOrigDataLen();
// NOTE: we handle zero sized data objects by "faking"
// a size of 1.
if (cost == 0) {
cost = 1;
}
string strResults="";
if (Trace::trace.getTraceType() == TRACE_TYPE_DEBUG2) {
strResults.append(id);
strResults.append("[I]=");
}
double utiltiy = getUtilityFunction()->compute(id, strResults);
HAGGLE_DBG2("%s --> %f\n", strResults.c_str(), utiltiy);
Timeval now = Timeval::now();
bool purgeNow = false;
if (utiltiy < getGlobalOptimizer()->getMinimumThreshold()) {
HAGGLE_DBG("Minimum threshold for incoming data object %s is insufficient: %f < %f\n", id.c_str(), utiltiy, getGlobalOptimizer()->getMinimumThreshold());
current_drop_on_insert++;
purgeNow = true;
}
if (!purgeOnInsert && ((current_size + cost) > max_capacity_kb*1024)) {
HAGGLE_DBG("Cache is full and purge on insert is disabled, evicting new data object!\n");
purgeNow = true;
total_do_hard_evicted++;
total_do_hard_evicted_bytes += cost;
}
if (purgeNow) {
// we need to properly remove from bloomfilter even when capacity is violated
getUtilityFunction()->notifyDelete(dObj);
// CBMEN, HL - Begin
// Remove any pending send events for this data object
HAGGLE_STAT("purging send events for dObj %s\n", dObj->getIdStr());
getManager()->getKernel()->cancelEvents(EVENT_TYPE_DATAOBJECT_SEND, dObj);
// CBMEN, HL, End
// delayed delete (although never inserted)
if (!keep_in_bloomfilter) {
int delay = (bloomfilter_remove_delay_ms <= 0) ? 1000 : bloomfilter_remove_delay_ms;
DataObjectId_t *heapId = (DataObjectId_t *)malloc(sizeof(DataObjectId_t));
DataObject::idStrToId(id, *heapId);
getManager()->getKernel()->addEvent(new Event(bloomfilterRemoveDelayEventType, heapId, delay/(double)1000));
}
return;
}
DataObjectUtilityMetadata *do_metadata = new DataObjectUtilityMetadata(
dObj,
id,
cost,
utiltiy,
now,
dObj->getCreateTime());
if (purgeOnInsert) {
do_metadata->setEnableDeletion(false);
}
if (false == utilMetadata.insert(make_pair(id, do_metadata)).second) {
HAGGLE_ERR("Somehow data object already in cache\n");
delete do_metadata;
return;
}
current_size += cost;
current_num_do++;
total_do_inserted++;
total_do_inserted_bytes += cost;
if (!purgeOnInsert) {
getManager()->insertDataObjectIntoDataStore(dObj);
return;
}
bool was_deleted = false;
_purgeCache(id, &was_deleted);
if (was_deleted) {
HAGGLE_DBG("Purged incoming data object %s on insert.\n", id.c_str());
current_drop_on_insert++;
return;
}
// DO still in cache, mark to allow deletion in future
do_metadata->setEnableDeletion(true);
getManager()->insertDataObjectIntoDataStore(dObj);
}