WARNING: This is the _old_ Lustre wiki, and it is in the process of being retired. The information found here is all likely to be out of date. Please search the new wiki for more up to date information.

Difference between revisions of "Clustered Metadata"

From Obsolete Lustre Wiki
Jump to navigationJump to search
m (Added link to HPCS CMD Design Doc PDF)
 
(28 intermediate revisions by 3 users not shown)
Line 1: Line 1:
= Clustered Metadata Design (in progress) =
+
<small>''(Updated: Sep 2010)''</small>
 +
__TOC__
 +
This document describes the design of the clustered metadata handling
 +
for Lustre™.  This material depends on other Lustre designs, such as:
  
This document describes the design of the clustered metadata handling
 
for Lustre.  This material depends on other Lustre design, such as:
 
:
 
 
* General recovery
 
* General recovery
 
* Orphan Recovery
 
* Orphan Recovery
 
* Metadata Write Back caching
 
* Metadata Write Back caching
  
The draft Clustered Metadata Design is available as a PDF:
+
For a draft of the design document, see [[Media:HPCS_CMD_06_15_09.pdf|''Clustered Metadata Design'']].
 
 
* [[Media:HPCS_CMD_06_15_09.pdf|Clustered Metadata Design]]
 
  
 
== Introduction ==
 
== Introduction ==
  
Overall the clustered metadata handling is structured as follows.  
+
Overall, the clustered metadata handling is structured as follows:  
  
* A cluster of metadata servers manage a collection of inode groups.   Each inode group is a Lustre device exporting the usual metadata   api, augmented with a few operations specifically crafted for   metadata clustering.  We call these collections of inodes inode   groups.
+
* A cluster of metadata servers manage a collection of inode groups. Each inode group is a Lustre device exporting the usual metadata API augmented with a few operations specifically crafted for metadata clustering.  We call these collections of inodes inode groups.
* Directory formats for file systems used on the MDS devices are   changed to introduce a allow directory entries to contain an inode   group and identifier of the inode.
+
* Directory formats for file systems used on the MDS devices are changed to allow directory entries to contain an inode group and identifier of the inode.
* A logical clustered metadata driver is introduced below the client   Lustre file system write back cache driver that maintains   connections with the MDS servers.
+
* A logical clustered metadata driver is introduced below the client Lustre file system write back cache driver that maintains connections with the MDS servers.
* There is a single metadata protocol that is used by the client file   system to make updates on the MDS's and by the MDS's to make   updates involving other MDS's.
+
* A single metadata protocol is used by the client file system to make updates on the MDSs and by the MDSs to make updates involving other MDSs.
* There is a single recovery protocol that is used by the clients -   MDS and MDS-MDS service.
+
* A single recovery protocol is used by the clients - MDS and MDS-MDS service.
* Directories can be split across multiple MDS nodes.  In that case a   primary MDS directory inode contains an extended attribute that   points at other MDS inodes which we call directory objects.
+
* Directories can be split across multiple MDS nodes.  In this case, a primary MDS directory inode contains an extended attribute that points at other MDS inodes, which we call directory objects.
  
== Configuration management and Startup ==
+
== Configuration management and startup ==
  
 
The configuration will name an MDS server, and optionally a failover
 
The configuration will name an MDS server, and optionally a failover
Line 30: Line 28:
  
 
They will also fetch from it a clustering descriptor.  The clustering
 
They will also fetch from it a clustering descriptor.  The clustering
descriptor contains a header and an array lists what inode groups are
+
descriptor contains a header, and an array lists which inode groups are
served by what server.
+
served by which server.
  
Through normal mechanisms clients will wait and probe for available
+
Through normal mechanisms, clients will wait and probe for available
 
metadata servers, during startup and cluster transitions.  When new
 
metadata servers, during startup and cluster transitions.  When new
servers are found or configurations have changed they can update their
+
servers are found or configurations have changed, they can update their
 
clustering descriptor as they update the LOV striping descriptor for
 
clustering descriptor as they update the LOV striping descriptor for
OST's.
+
OSTs.
  
 
== Data Structures ==
 
== Data Structures ==
  
The fid will contain a new 32 bit integer to name the inode group.   
+
The ''fid'' contains a new 32 bit integer to name the inode group.   
 
 
Directory entries will contain a new 32 bit integer to name the inode
 
group.
 
  
 
Directory inodes on the MDS, when large, contain a new EA which is a
 
Directory inodes on the MDS, when large, contain a new EA which is a
 
descriptor of how the directory is split over directory objects,
 
descriptor of how the directory is split over directory objects,
residing on other MDS's.  This EA is subject to ordinary concurrency
+
residing on other MDSs.  This EA is subject to ordinary concurrency
 
control by the MDS holding the inode.  The EA is virtually identical
 
control by the MDS holding the inode.  The EA is virtually identical
to the LOV EA.  
+
to the LOV EA.
  
 
== The clustered metadata client (CMC) ==
 
== The clustered metadata client (CMC) ==
  
Client systems will have the write back client (WBD) or client file
+
The function of the CMC is to figure out from the command issued which MDC to use.  This is based on:  
system directly communicate with the CMC driver: it offers the
+
* The inode groups in the request
metadata api to the file system and uses the metadata api offered by a
+
* A hash value of names used in the request, combined with the EA of a primary inode involved in the request
collection of MDC drivers.  Each MDC driver managed the metadata
+
* For ''readdir'', the directory offset combined with the EA of the primary inode
traffic to one.
+
* The clustering descriptor
 
 
The function of the CMC is very simple: it figures out from the
 
command issued what MDC to use.  This is based on:  
 
* the inode groups in the request
 
* a hash value of names used in the request, combined with the EA of   a primary inode involved in the request.
 
* for readdir the directory offset combined with the EA of the   primary inode
 
* the clustering descriptor
 
  
In any case every command is dispatched to a single metadata server,
+
In any case, every command is dispatched to a single metadata server and
 
the clients will not engage more than one metadata server for a single
 
the clients will not engage more than one metadata server for a single
 
request.   
 
request.   
  
The api changes here are minimal and the client part of the
+
The API changes here are minimal and the client part of the implementation is trivial.
implementation is very trivial.
 
  
 
== MDS implementation ==
 
== MDS implementation ==
  
For the most part, operations are extremely similar or identical to
+
For the most part, operations are similar or identical to what they were before. In some cases, multiple MDS servers are involved in updates.
what they were before.   In some cases multiple mds servers are
 
involved in updates.
 
  
Getattr, open, readdir, setattr and lookup methods are unaffected.  
+
''getattr'', ''open'', ''readdir'', ''setattr'' and ''lookup'' methods are unaffected.  
  
 
Methods adding entries to directories are modified in some cases:  
 
Methods adding entries to directories are modified in some cases:  
  
* '''mkdir''' always create the new directory on another MDS
+
* ''mkdir'' always creates the new directory on another MDS.
* '''unlink, rmdir, rename''': may involve more than one MDS
+
* ''unlink'', ''rmdir'', and ''rename'' may involve more than one MDS.
* '''large directories''' all operations making updates to   directories can cause a directory split. The directory split is  discussed below.
+
* For ''large directories'', all operations making updates to directories can cause a [[#Directory_Split|directory split]].
* '''other operations''' If no splits large directories are   encountered all other operations proceed as they are executed on   one MDS.
+
* For ''other operations'', if no splits in large directories are encountered, all other operations proceed as they are executed on one MDS.
  
 
=== Directory Split ===
 
=== Directory Split ===
  
A directory that is growing larger will be split. There is a fairly heavy penalty associated with splitting the directory and also with renames in within split directories. Moreover, at the point of splitting, inodes become remote and will incur a penalty upon unlink.
+
A directory can be striped over several MDTs as files over several OSTs. Then the directory will be split into several objects and each one will be located in different MDTs. The layout information(stripe EA) will be stored in the extend attributes of all split objects.
 
 
Probably it is best to delay the split until the directory is fairly large, and then to split over several nodes, to avoid further splits being necessary soon afterwards.
 
  
 
== Recovery ==
 
== Recovery ==
  
=== Transaction Replay ===
+
Initially, metadata operations that span multiple MDSs (MDTs) will be ordered and  synchronous to simplify recovery from a system crash. This may impact the performance of operations involving several MDTsAlso, an inode leak may occur after MDS recovery, but only in such a way that data is never lost.  These leaked inodes will be deleted by lfsck verification of the MDT filesystems.
 
 
The MDS - MDS interaction is managed as follows.  The node approached
 
with a request change is made the coordinator of the transaction.
 
 
 
==== Mechanisms ====
 
 
 
The coordinator will first establish that the transaction can commit on all nodes, by acquiring locks on directories and checking for available space existing entries with the same name etc.  It may also first perform a directory split if the size is becoming too large, and more MDS nodes are still available. 
 
 
 
All nodes involved in the transaction need to have a transaction sequence number to place the transaction into their sequence and allow correctly replay.
 
 
 
At this point the coordinator will:
 
* start a transaction locally.
 
* It will then report the transaction sequence number to all other nodes involved in the transaction. 
 
* These nodes will commit (in memory as usual), write a journal record for replay and reply to the coordinator. 
 
* The coordinator will then commit its own transaction.
 
* The replay log records are subject to normal log commit cancelation messages, but on the coordinator commit messages must be received from all other nodes before the record will be canceled.
 
 
 
In this way if the results of the transaction survive on any of the nodes, they can be replayed on all.
 
 
 
==== Cluster crashes and the transaction sequence ====
 
 
 
If the cluster crashes abruptly, there is the opportunity for transactions to be in progress affecting multiple nodes. Dependencies between the transactions must be managed to ensure serializibility of the protocol.
 
 
 
 
 
Example: In transaction one, a node X creates directories a. Then in transaction 2 a cross MDS node rename moves a file with a directory entry on node Y into this directory.  It is now possible for this file to lose its directory entry on Y and for the transaction on X not to commit. More complex examples exist.
 
 
 
 
 
We see 4 solutions:
 
 
 
* Disk commit delay locks: dependent transactions many not commit before the parent transaction commits.
 
* Commit acks: transactions may not proceed until previous pre-requisite transactions have committed.
 
* Synchronous NVRAM journal on all MDS nodes
 
* Shared journal among all MDS nodes
 
 
 
The first and last method offer the most opportunity to proceed without synchronous disk writes. The last method involves contention on a shared resource.
 
 
 
Although exhaustive analysis remains, it is clear that rename and splitting the directories are the primary culprits. Hence, we wonder about the following policy related issues:
 
* Only split really large directories, say after 1M entries.
 
* Do not needlessly create subdirectories on other nodes.  A much better policy is likely to keep directories with one owner, or possibly one client system generating them together.
 
 
 
 
 
==== Replay ====
 
 
 
 
 
To order transaction sequences Lustre uses reply ack's: the acks
 
server only one purpose to release a lock that enforces ordering of
 
the transaction sequence.  In the case where MDS operations involve
 
more than server, the reply "ack" from the primary to secondary
 
servers should only be sent after the client has sent the ack to the
 
first server.  This MDS-MDS reply ack is now not really an ack anymore
 
but a simple lock cancelation review.
 
 
 
Clients will replay lost transactions to the mds which they originally
 
engaged for the request.   
 
 
 
 
 
Orphaned children will be cleaned up only
 
after replay completes to allow orphaned objects to be re-used during
 
replay.
 
 
 
 
 
=== Failover ===
 
 
 
The configuration data can designate a standby MDS that will take over
 
from a failed MDS.  By organizing the servers in one or more rings,
 
the nearest working left neighbor MDS can be the failover node.  This
 
leads to a simple scheme with multiple failover nodes, avoiding quorum
 
and other complications beyond what is needed for two node clusters.
 
  
 +
In the long term, CMD recovery will rely on global epochs, which will allow distributed asynchronous updates to multiple MDTs.
  
 
== Locking ==
 
== Locking ==
  
We believe locking can be done in fid order as it is currently done on the MDS.
+
We believe locking can be done in ''fid'' order as it is currently done on the MDS.

Latest revision as of 11:36, 10 September 2010

(Updated: Sep 2010)

This document describes the design of the clustered metadata handling for Lustre™. This material depends on other Lustre designs, such as:

  • General recovery
  • Orphan Recovery
  • Metadata Write Back caching

For a draft of the design document, see Clustered Metadata Design.

Introduction

Overall, the clustered metadata handling is structured as follows:

  • A cluster of metadata servers manage a collection of inode groups. Each inode group is a Lustre device exporting the usual metadata API augmented with a few operations specifically crafted for metadata clustering. We call these collections of inodes inode groups.
  • Directory formats for file systems used on the MDS devices are changed to allow directory entries to contain an inode group and identifier of the inode.
  • A logical clustered metadata driver is introduced below the client Lustre file system write back cache driver that maintains connections with the MDS servers.
  • A single metadata protocol is used by the client file system to make updates on the MDSs and by the MDSs to make updates involving other MDSs.
  • A single recovery protocol is used by the clients - MDS and MDS-MDS service.
  • Directories can be split across multiple MDS nodes. In this case, a primary MDS directory inode contains an extended attribute that points at other MDS inodes, which we call directory objects.

Configuration management and startup

The configuration will name an MDS server, and optionally a failover node, which hold the root inode for a fileset. Clients will contact that MDS for the root inode during mount, as they do already.

They will also fetch from it a clustering descriptor. The clustering descriptor contains a header, and an array lists which inode groups are served by which server.

Through normal mechanisms, clients will wait and probe for available metadata servers, during startup and cluster transitions. When new servers are found or configurations have changed, they can update their clustering descriptor as they update the LOV striping descriptor for OSTs.

Data Structures

The fid contains a new 32 bit integer to name the inode group.

Directory inodes on the MDS, when large, contain a new EA which is a descriptor of how the directory is split over directory objects, residing on other MDSs. This EA is subject to ordinary concurrency control by the MDS holding the inode. The EA is virtually identical to the LOV EA.

The clustered metadata client (CMC)

The function of the CMC is to figure out from the command issued which MDC to use. This is based on:

  • The inode groups in the request
  • A hash value of names used in the request, combined with the EA of a primary inode involved in the request
  • For readdir, the directory offset combined with the EA of the primary inode
  • The clustering descriptor

In any case, every command is dispatched to a single metadata server and the clients will not engage more than one metadata server for a single request.

The API changes here are minimal and the client part of the implementation is trivial.

MDS implementation

For the most part, operations are similar or identical to what they were before. In some cases, multiple MDS servers are involved in updates.

getattr, open, readdir, setattr and lookup methods are unaffected.

Methods adding entries to directories are modified in some cases:

  • mkdir always creates the new directory on another MDS.
  • unlink, rmdir, and rename may involve more than one MDS.
  • For large directories, all operations making updates to directories can cause a directory split.
  • For other operations, if no splits in large directories are encountered, all other operations proceed as they are executed on one MDS.

Directory Split

A directory can be striped over several MDTs as files over several OSTs. Then the directory will be split into several objects and each one will be located in different MDTs. The layout information(stripe EA) will be stored in the extend attributes of all split objects.

Recovery

Initially, metadata operations that span multiple MDSs (MDTs) will be ordered and synchronous to simplify recovery from a system crash. This may impact the performance of operations involving several MDTs. Also, an inode leak may occur after MDS recovery, but only in such a way that data is never lost. These leaked inodes will be deleted by lfsck verification of the MDT filesystems.

In the long term, CMD recovery will rely on global epochs, which will allow distributed asynchronous updates to multiple MDTs.

Locking

We believe locking can be done in fid order as it is currently done on the MDS.