CernVM-FS  2.9.0
 All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Macros Pages
GarbageCollector< CatalogTraversalT, HashFilterT > Class Template Reference

#include <garbage_collector.h>

Collaboration diagram for GarbageCollector< CatalogTraversalT, HashFilterT >:

Classes

struct  Configuration
 
class  ReflogBasedInfoShim
 

Public Member Functions

 GarbageCollector (const Configuration &configuration)
 
void UseReflogTimestamps ()
 
bool Collect ()
 
unsigned int preserved_catalog_count () const
 
unsigned int condemned_catalog_count () const
 
unsigned int condemned_objects_count () const
 
uint64_t condemned_bytes_count () const
 
uint64_t oldest_trunk_catalog () const
 

Protected Types

typedef
CatalogTraversalT::ObjectFetcherTN 
ObjectFetcherTN
 
typedef ObjectFetcherTN::HistoryTN HistoryTN
 
typedef ObjectFetcherTN::ReflogTN ReflogTN
 
typedef
CatalogTraversalT::CatalogTN 
CatalogTN
 
typedef
CatalogTraversalT::CallbackDataTN 
TraversalCallbackDataTN
 
typedef
CatalogTraversalT::Parameters 
TraversalParameters
 
typedef std::vector< shash::AnyHashVector
 

Protected Member Functions

TraversalParameters GetTraversalParams (const Configuration &configuration)
 
void PreserveDataObjects (const TraversalCallbackDataTN &data)
 
void SweepDataObjects (const TraversalCallbackDataTN &data)
 
bool AnalyzePreservedCatalogTree ()
 
bool CheckPreservedRevisions ()
 
bool SweepReflog ()
 
void CheckAndSweep (const shash::Any &hash)
 
void Sweep (const shash::Any &hash)
 
bool RemoveCatalogFromReflog (const shash::Any &catalog)
 
void PrintCatalogTreeEntry (const unsigned int tree_level, const CatalogTN *catalog) const
 
void LogDeletion (const shash::Any &hash) const
 

Private Attributes

const Configuration configuration_
 
ReflogBasedInfoShim catalog_info_shim_
 
CatalogTraversalT traversal_
 
HashFilterT hash_filter_
 
bool use_reflog_timestamps_
 
uint64_t oldest_trunk_catalog_
 
bool oldest_trunk_catalog_found_
 
unsigned int preserved_catalogs_
 
unsigned int unreferenced_trees_
 
unsigned int condemned_trees_
 
unsigned int condemned_catalogs_
 
float last_reported_status_
 
unsigned int condemned_objects_
 
uint64_t condemned_bytes_
 

Detailed Description

template<class CatalogTraversalT, class HashFilterT>
class GarbageCollector< CatalogTraversalT, HashFilterT >

This file is part of the CernVM File System.

The GarbageCollector class is figuring out which data objects (represented by their content hashes) can be deleted as outdated garbage. Garbage collection is performed on the granularity of catalog revisions, thus a complete repository revision is either considered to be outdated or active. This way, a mountable repository revision stays completely usable (no nested catalogs or data objects become unavailable). A revision is defined by it's root catalog; the revision numbers of nested catalogs are irrelevant, since they might be referenced by newer (preserved) repository revisions. Thus, garbage objects are those that are not referenced by any of the pre- served root catalogs or their direct subordinate nested catalog try.

We use a two-stage approach:

1st Stage - Initialization The GarbageCollector is reading all the catalogs that are meant to be preserved. It builds up a filter (HashFilterT) containing all content hashes that are not to be deleted

2nd Stage - Sweeping The initialized HashFilterT is presented with all content hashes found in condemned catalogs and decides if they are referenced by the preserved catalog revisions or not.

The GarbageCollector is templated with CatalogTraversalT mainly for testability and with HashFilterT as an instance of the Strategy Pattern to abstract from the actual hash filtering method to be used.

Definition at line 45 of file garbage_collector.h.

Member Typedef Documentation

template<class CatalogTraversalT, class HashFilterT>
typedef CatalogTraversalT::CatalogTN GarbageCollector< CatalogTraversalT, HashFilterT >::CatalogTN
protected

Definition at line 50 of file garbage_collector.h.

template<class CatalogTraversalT, class HashFilterT>
typedef std::vector<shash::Any> GarbageCollector< CatalogTraversalT, HashFilterT >::HashVector
protected

Definition at line 53 of file garbage_collector.h.

template<class CatalogTraversalT, class HashFilterT>
typedef ObjectFetcherTN::HistoryTN GarbageCollector< CatalogTraversalT, HashFilterT >::HistoryTN
protected

Definition at line 48 of file garbage_collector.h.

template<class CatalogTraversalT, class HashFilterT>
typedef CatalogTraversalT::ObjectFetcherTN GarbageCollector< CatalogTraversalT, HashFilterT >::ObjectFetcherTN
protected

Definition at line 47 of file garbage_collector.h.

template<class CatalogTraversalT, class HashFilterT>
typedef ObjectFetcherTN::ReflogTN GarbageCollector< CatalogTraversalT, HashFilterT >::ReflogTN
protected

Definition at line 49 of file garbage_collector.h.

template<class CatalogTraversalT, class HashFilterT>
typedef CatalogTraversalT::CallbackDataTN GarbageCollector< CatalogTraversalT, HashFilterT >::TraversalCallbackDataTN
protected

Definition at line 51 of file garbage_collector.h.

template<class CatalogTraversalT, class HashFilterT>
typedef CatalogTraversalT::Parameters GarbageCollector< CatalogTraversalT, HashFilterT >::TraversalParameters
protected

Definition at line 52 of file garbage_collector.h.

Constructor & Destructor Documentation

template<class CatalogTraversalT , class HashFilterT >
GarbageCollector< CatalogTraversalT, HashFilterT >::GarbageCollector ( const Configuration configuration)
explicit

Definition at line 31 of file garbage_collector_impl.h.

Here is the call graph for this function:

Member Function Documentation

template<class CatalogTraversalT , class HashFilterT >
bool GarbageCollector< CatalogTraversalT, HashFilterT >::AnalyzePreservedCatalogTree ( )
protected

Definition at line 217 of file garbage_collector_impl.h.

Here is the call graph for this function:

template<class CatalogTraversalT , class HashFilterT >
void GarbageCollector< CatalogTraversalT, HashFilterT >::CheckAndSweep ( const shash::Any hash)
protected

Definition at line 166 of file garbage_collector_impl.h.

template<class CatalogTraversalT , class HashFilterT >
bool GarbageCollector< CatalogTraversalT, HashFilterT >::CheckPreservedRevisions ( )
protected

Definition at line 241 of file garbage_collector_impl.h.

template<class CatalogTraversalT , class HashFilterT >
bool GarbageCollector< CatalogTraversalT, HashFilterT >::Collect ( )

Definition at line 208 of file garbage_collector_impl.h.

Referenced by swissknife::CommandGc::Main().

Here is the caller graph for this function:

template<class CatalogTraversalT, class HashFilterT>
uint64_t GarbageCollector< CatalogTraversalT, HashFilterT >::condemned_bytes_count ( ) const
inline

Definition at line 99 of file garbage_collector.h.

template<class CatalogTraversalT, class HashFilterT>
unsigned int GarbageCollector< CatalogTraversalT, HashFilterT >::condemned_catalog_count ( ) const
inline

Definition at line 97 of file garbage_collector.h.

template<class CatalogTraversalT, class HashFilterT>
unsigned int GarbageCollector< CatalogTraversalT, HashFilterT >::condemned_objects_count ( ) const
inline

Definition at line 98 of file garbage_collector.h.

template<class CatalogTraversalT, class HashFilterT>
GarbageCollector< CatalogTraversalT, HashFilterT >::TraversalParameters GarbageCollector< CatalogTraversalT, HashFilterT >::GetTraversalParams ( const Configuration configuration)
protected

Definition at line 63 of file garbage_collector_impl.h.

template<class CatalogTraversalT , class HashFilterT >
void GarbageCollector< CatalogTraversalT, HashFilterT >::LogDeletion ( const shash::Any hash) const
protected

Definition at line 341 of file garbage_collector_impl.h.

Here is the call graph for this function:

template<class CatalogTraversalT, class HashFilterT>
uint64_t GarbageCollector< CatalogTraversalT, HashFilterT >::oldest_trunk_catalog ( ) const
inline

Definition at line 100 of file garbage_collector.h.

Referenced by swissknife::CommandGc::Main().

Here is the caller graph for this function:

template<class CatalogTraversalT, class HashFilterT>
unsigned int GarbageCollector< CatalogTraversalT, HashFilterT >::preserved_catalog_count ( ) const
inline

Definition at line 96 of file garbage_collector.h.

template<class CatalogTraversalT, class HashFilterT>
void GarbageCollector< CatalogTraversalT, HashFilterT >::PreserveDataObjects ( const TraversalCallbackDataTN data)
protected

Definition at line 79 of file garbage_collector_impl.h.

Here is the call graph for this function:

template<class CatalogTraversalT , class HashFilterT >
void GarbageCollector< CatalogTraversalT, HashFilterT >::PrintCatalogTreeEntry ( const unsigned int  tree_level,
const CatalogTN catalog 
) const
protected

Definition at line 317 of file garbage_collector_impl.h.

template<class CatalogTraversalT , class HashFilterT >
bool GarbageCollector< CatalogTraversalT, HashFilterT >::RemoveCatalogFromReflog ( const shash::Any catalog)
protected

Definition at line 198 of file garbage_collector_impl.h.

Here is the call graph for this function:

template<class CatalogTraversalT , class HashFilterT >
void GarbageCollector< CatalogTraversalT, HashFilterT >::Sweep ( const shash::Any hash)
protected

Definition at line 175 of file garbage_collector_impl.h.

Here is the call graph for this function:

template<class CatalogTraversalT, class HashFilterT>
void GarbageCollector< CatalogTraversalT, HashFilterT >::SweepDataObjects ( const TraversalCallbackDataTN data)
protected

Definition at line 122 of file garbage_collector_impl.h.

Here is the call graph for this function:

template<class CatalogTraversalT , class HashFilterT >
bool GarbageCollector< CatalogTraversalT, HashFilterT >::SweepReflog ( )
protected

Definition at line 254 of file garbage_collector_impl.h.

Here is the call graph for this function:

template<class CatalogTraversalT , class HashFilterT >
void GarbageCollector< CatalogTraversalT, HashFilterT >::UseReflogTimestamps ( )

Definition at line 55 of file garbage_collector_impl.h.

Referenced by swissknife::CommandGc::Main().

Here is the caller graph for this function:

Member Data Documentation

template<class CatalogTraversalT, class HashFilterT>
ReflogBasedInfoShim GarbageCollector< CatalogTraversalT, HashFilterT >::catalog_info_shim_
private

Definition at line 144 of file garbage_collector.h.

template<class CatalogTraversalT, class HashFilterT>
uint64_t GarbageCollector< CatalogTraversalT, HashFilterT >::condemned_bytes_
private
template<class CatalogTraversalT, class HashFilterT>
unsigned int GarbageCollector< CatalogTraversalT, HashFilterT >::condemned_catalogs_
private

Number of catalogs garbage collected, count grows as GC progresses

Definition at line 170 of file garbage_collector.h.

Referenced by GarbageCollector< CatalogTraversalT, HashFilterT >::condemned_catalog_count().

template<class CatalogTraversalT, class HashFilterT>
unsigned int GarbageCollector< CatalogTraversalT, HashFilterT >::condemned_objects_
private
template<class CatalogTraversalT, class HashFilterT>
unsigned int GarbageCollector< CatalogTraversalT, HashFilterT >::condemned_trees_
private

Number of root catalogs garbage collected, count grows as GC progresses

Definition at line 166 of file garbage_collector.h.

template<class CatalogTraversalT, class HashFilterT>
const Configuration GarbageCollector< CatalogTraversalT, HashFilterT >::configuration_
private
template<class CatalogTraversalT, class HashFilterT>
HashFilterT GarbageCollector< CatalogTraversalT, HashFilterT >::hash_filter_
private

Definition at line 146 of file garbage_collector.h.

template<class CatalogTraversalT, class HashFilterT>
float GarbageCollector< CatalogTraversalT, HashFilterT >::last_reported_status_
private

Keeps track if the last status report issued, between 0 and 1

Definition at line 174 of file garbage_collector.h.

template<class CatalogTraversalT, class HashFilterT>
uint64_t GarbageCollector< CatalogTraversalT, HashFilterT >::oldest_trunk_catalog_
private

A marker for the garbage collection grace period, the time span that is walked back from the current head catalog. There can be named snapshots older than this snapshot. The oldest_trunk_catalog_ is used as a marker for when to remove auxiliary files (meta info, history, ...).

Definition at line 155 of file garbage_collector.h.

Referenced by GarbageCollector< CatalogTraversalT, HashFilterT >::oldest_trunk_catalog().

template<class CatalogTraversalT, class HashFilterT>
bool GarbageCollector< CatalogTraversalT, HashFilterT >::oldest_trunk_catalog_found_
private

Definition at line 156 of file garbage_collector.h.

template<class CatalogTraversalT, class HashFilterT>
unsigned int GarbageCollector< CatalogTraversalT, HashFilterT >::preserved_catalogs_
private
template<class CatalogTraversalT, class HashFilterT>
CatalogTraversalT GarbageCollector< CatalogTraversalT, HashFilterT >::traversal_
private

Definition at line 145 of file garbage_collector.h.

template<class CatalogTraversalT, class HashFilterT>
unsigned int GarbageCollector< CatalogTraversalT, HashFilterT >::unreferenced_trees_
private

Number of catalogs in the reflog that are to be deleted (in fact, some of them might not exist anymore).

Definition at line 162 of file garbage_collector.h.

template<class CatalogTraversalT, class HashFilterT>
bool GarbageCollector< CatalogTraversalT, HashFilterT >::use_reflog_timestamps_
private

Definition at line 148 of file garbage_collector.h.


The documentation for this class was generated from the following files: