Ceph Day London 2014 - The current state of CephFS development

CephFS Update

John [email protected]

Ceph Day London

mailto:[email protected]

2 Ceph Day London - CephFS Update

Agenda

● Introduction to distributed filesystems

● Architectural overview

● Recent development

● Test & QA

3Ceph Day London – CephFS Update

Distributed filesystems...and why they are hard.


Interfaces to storage

● Object● Ceph RGW, S3, Swift

● Block (aka SAN)● Ceph RBD, iSCSI, FC, SAS

● File (aka scale-out NAS)● Ceph, GlusterFS, Lustre, proprietary filers


Interfaces to storage

FILE SYSTEMCephFS

BLOCK STORAGE

RBD

OBJECT STORAGE

RGW

Keystone

Geo-Replication

Native API

Multi-tenant

S3 & Swift

OpenStack

Linux Kernel

iSCSI

Clones

Snapshots

CIFS/NFS

HDFS

Distributed Metadata

Linux Kernel

POSIX


Object stores scale out well

● Last writer wins consistency

● Consistency rules only apply to one object at a time

● Clients are stateless (unless explicitly doing lock ops)

● No relationships exist between objects

● Objects have exactly one name

● Scale-out accomplished by mapping objects to nodes

● Single objects may be lost without affecting others


POSIX filesystems are hard to scale out

● Extents written from multiple clients must win or lose on all-or-nothing basis → locking

● Inodes depend on one another (directory hierarchy)

● Clients are stateful: holding files open

● Users have local-filesystem latency expectations: applications assume FS client will do lots of metadata caching for them.

● Scale-out requires spanning inode/dentry relationships across servers

● Loss of data can damage whole subtrees


Failure cases increase complexity further

● What should we do when... ?● Filesystem is full● Client goes dark● An MDS goes dark● Memory is running low● Clients are competing for the same files● Clients misbehave

● Hard problems in distributed systems generally, especially hard when we have to uphold POSIX semantics designed for local systems.


Terminology

● inode: a file. Has unique ID, may be referenced by one or more dentries.

● dentry: a link between an inode and a directory

● directory: special type of inode that has 0 or more child dentries

● hard link: many dentries referring to the same inode

● Terms originate form original (local disk) filesystems, where these were how a filesystem was represented on disk.


Architectural overview


CephFS architecture

● Dynamically balanced scale-out metadata

● Inherit flexibility/scalability of RADOS for data

● POSIX compatibility

● Beyond POSIX: Subtree snapshots, recursive statistics

Weil, Sage A., et al. "Ceph: A scalable, high-performance distributed file system." Proceedings of the 7th symposium on Operating systems

design and implementation. USENIX Association, 2006.http://ceph.com/papers/weil-ceph-osdi06.pdf

http://ceph.com/papers/weil-ceph-osdi06.pdf


Components

● Client: kernel, fuse, libcephfs● Server: MDS daemon● Storage: RADOS cluster (mons & OSDs)


Components

Linux host

M M

M

Ceph server daemons

ceph.ko

datametadata 0110


From application to disk

ceph-mds

libcephfsceph-fuse Kernel client

RADOS

Client network protocol

Application

Disk


Scaling out FS metadata

● Options for distributing metadata?– by static subvolume

– by path hash

– by dynamic subtree

● Consider performance, ease of implementation


DYNAMIC SUBTREE PARTITIONING


Dynamic subtree placement

● Locality: get the dentries in a dir from one MDS

● Support read heavy workloads by replicating non-authoritative copies (cached with capabilities just like clients do)

● In practice work at directory fragment level in order to handle large dirs


Data placement

● Stripe file contents across RADOS objects● get full rados cluster bandwidth from clients● delegate all placement/balancing to RADOS

● Control striping with layout vxattrs● layouts also select between multiple data pools

● Deletion is a special case: client deletions mark files 'stray', RADOS delete ops sent by MDS


Clients

● Two implementations:● ceph-fuse/libcephfs● kclient

● Interplay with VFS page cache, efficiency harder with fuse (extraneous stats etc)

● Client perf. matters, for single-client workloads

● Slow client can hold up others if it's hogging metadata locks: include clients in troubleshooting

● - future: want more per client perf stats and maybe metadata QoS per client. Clients probably group into jobs or workloads.

● - future: may want to tag client io with job id (eg hpc workload, samba client I'd, container/VM id)


Journaling and caching in MDS

● Metadata ops initially journaled to striped journal "file" in the metadata pool.

● I/O latency on metadata ops is sum of network latency and journal commit latency.

● Metadata remains pinned in in-memory cache until expired from journal.


Journaling and caching in MDS

● In some workloads we expect almost all metadata always in cache, in others its more of a stream.

● Control cache size with mds_cache_size

● Cache eviction relies on client cooperation

● MDS journal replay not only recovers data but also warms up cache. Use standby replay to keep that cache warm.


Lookup by inode

● Sometimes we need inode → path mapping:● Hard links● NFS handles

● Costly to store this: mitigate by piggybacking paths (backtraces) onto data objects

● Con: storing metadata to data pool● Con: extra IOs to set backtraces● Pro: disaster recovery from data pool

● Future: improve backtrace writing latency?


Extra features

● Snapshots:● Exploit RADOS snapshotting for file data● … plus some clever code in the MDS● Fast petabyte snapshots

● Recursive statistics● Lazily updated● Access via vxattr● Avoid spurious client I/O for df


CephFS in practice

ceph-deploy mds create myserver

ceph osd pool create fs_data

ceph osd pool create fs_metadata

ceph fs new myfs fs_metadata fs_data

mount -t cephfs x.x.x.x:6789 /mnt/ceph


Managing CephFS clients

● New in giant: see hostnames of connected clients

● Client eviction is sometimes important:● Skip the wait during reconnect phase on MDS restart● Allow others to access files locked by crashed client

● Use OpTracker to inspect ongoing operations


CephFS tips

● Choose MDS servers with lots of RAM

● Investigate clients when diagnosing stuck/slow access

● Use recent Ceph and recent kernel

● Use a conservative configuration:● Single active MDS, plus one standby● Dedicated MDS server● Kernel client● No snapshots, no inline data


Development update


RADOS

A reliable, autonomous, distributed object store comprised of self-healing, self-managing, intelligent storage nodes

RADOS

A reliable, autonomous, distributed object store comprised of self-healing, self-managing, intelligent storage nodes

LIBRADOS

A library allowingapps to directlyaccess RADOS,with support forC, C++, Java,Python, Ruby,and PHP

LIBRADOS

A library allowingapps to directlyaccess RADOS,with support forC, C++, Java,Python, Ruby,and PHP

RBD

A reliable and fully-distributed block device, with a Linux kernel client and a QEMU/KVM driver

RBD

A reliable and fully-distributed block device, with a Linux kernel client and a QEMU/KVM driver

RADOSGW

A bucket-based REST gateway, compatible with S3 and Swift

RADOSGW

A bucket-based REST gateway, compatible with S3 and Swift

APPAPP APPAPP HOST/VMHOST/VM CLIENTCLIENT

CEPH FS

A POSIX-compliant distributed file system, with a Linux kernel client and support for FUSE

CEPH FS

A POSIX-compliant distributed file system, with a Linux kernel client and support for FUSE

NEARLYAWESOME

AWESOMEAWESOME

AWESOME

AWESOME


Towards a production-ready CephFS

● Focus on resilience:

1. Don't corrupt things

2. Stay up

3. Handle the corner cases

4. When something is wrong, tell me

5. Provide the tools to diagnose and fix problems

● Achieve this first within a conservative single-MDS configuration


Giant → Hammer timeframe

● Initial online fsck (a.k.a. forward scrub)

● Online diagnostics (`session ls`, MDS health alerts)

● Journal resilience & tools (cephfs-journal-tool)

● flock in the FUSE client

● Initial soft quota support

● General resilience: full OSDs, full metadata cache


FSCK and repair

● Recover from damage:● Loss of data objects (which files are damaged?)● Loss of metadata objects (what subtree is damaged?)

● Continuous verification:● Are recursive stats consistent?● Does metadata on disk match cache?● Does file size metadata match data on disk?

● Repair:● Automatic where possible● Manual tools to enable support


Client management

● Current eviction is not 100% safe against rogue clients● Update to client protocol to wait for OSD blacklist

● Client metadata● Initially domain name, mount point● Extension to other identifiers?


Online diagnostics

● Bugs exposed relate to failures of one client to release resources for another client: “my filesystem is frozen”. Introduce new health messages:

● “client xyz is failing to respond to cache pressure”● “client xyz is ignoring capability release messages”● Add client metadata to allow us to give domain names

instead of IP addrs in messages.

● Opaque behavior in the face of dead clients. Introduce `session ls`

● Which clients does MDS think are stale?● Identify clients to evict with `session evict`


Journal resilience

● Bad journal prevents MDS recovery: “my MDS crashes on startup”:

● Data loss● Software bugs

● Updated on-disk format to make recovery from damage easier

● New tool: cephfs-journal-tool● Inspect the journal, search/filter● Chop out unwanted entries/regions


Handling resource limits

● Write a test, see what breaks!

● Full MDS cache:● Require some free memory to make progress● Require client cooperation to unpin cache objects● Anticipate tuning required for cache behaviour: what

should we evict?

● Full OSD cluster● Require explicit handling to abort with -ENOSPC

● MDS → RADOS flow control:● Contention between I/O to flush cache and I/O to journal


Test, QA, bug fixes

● The answer to “Is CephFS production ready?”

● teuthology test framework:● Long running/thrashing test● Third party FS correctness tests● Python functional tests

● We dogfood CephFS internally● Various kclient fixes discovered● Motivation for new health monitoring metrics

● Third party testing is extremely valuable


What's next?

● You tell us!

● Recent survey highlighted:● FSCK hardening● Multi-MDS hardening● Quota support

● Which use cases will matter to community?● Backup● Hadoop● NFS/Samba gateway● Other?


Reporting bugs

● Does the most recent development release or kernel fix your issue?

● What is your configuration? MDS config, Ceph version, client version, kclient or fuse

● What is your workload?

● Can you reproduce with debug logging enabled?

http://ceph.com/resources/mailing-list-irc/

http://tracker.ceph.com/projects/ceph/issues

http://ceph.com/docs/master/rados/troubleshooting/log-and-debug/

http://ceph.com/resources/mailing-list-irc/

http://tracker.ceph.com/projects/ceph/issues

http://ceph.com/docs/master/rados/troubleshooting/log-and-debug/


Future

● Ceph Developer Summit:● When: 8 October● Where: online

● Post-Hammer work:● Recent survey highlighted multi-MDS, quota support ● Testing with clustered Samba/NFS?


Questions?


Body slide design guidelines

● > 15 words per bullet

● If your slide is text-only, reserve at least 1/3 of the slide for white space.

● If you use a graphic, make sure text is readable.


Introduce Red Hat

● Create an agenda slide for every presentation.● Outline what you’re going to tell the audience.● Prepare them for a call to action after the presentation.

● If this is a confidential presentation, use the confidential presentation template located on the Corporate > Templates > Presentation templates page of the PNT Portal.

https://engineering.redhat.com/pnt/p-412440/salestraining...-template.otp


Introduce Red Hat solutions and services

● Provide product details that specifically solve the customer pain point you’re addressing.

● These slides explain how Red Hat solutions work, what makes them unique and valuable.


Learn more

● End with a call to action.

● Let the audience know what can be done next, how you or Red Hat can help them.


Divider slide


Divider SlideDivider slide

A STORAGE REVOLUTION

PROPRIETARY HARDWARE

PROPRIETARY SOFTWARE

SUPPORT & MAINTENANCE

COMPUTER DISK

COMPUTER DISK

COMPUTER DISK

STANDARDHARDWARE

OPEN SOURCE SOFTWARE

ENTERPRISEPRODUCTS &

SERVICES

COMPUTER DISK

COMPUTER DISK

COMPUTER DISK

Copyright © 2014 by Inktank | Private and Confidential

ARCHITECTURAL COMPONENTS

58

RGWA web services

gateway for object storage, compatible

with S3 and Swift

LIBRADOSA library allowing apps to directly access RADOS (C, C++, Java, Python, Ruby, PHP)

RADOSA software-based, reliable, autonomous, distributed object store comprised ofself-healing, self-managing, intelligent storage nodes and lightweight monitors

RBDA reliable, fully-distributed block device with cloud

platform integration

CEPHFSA distributed file

system with POSIX semantics and scale-

out metadata management

APP HOST/VM CLIENT



59

RGWA web services


with S3 and Swift








APP HOST/VM CLIENT

OBJECT STORAGE DAEMONS

60

FS

DISK

OSD

DISK

OSD

FS

DISK

OSD

FS

DISK

OSD

FS

btrfsxfsext4

M

M

M

RADOS CLUSTER

61

APPLICATION

M M

M M

M

RADOS CLUSTER

RADOS COMPONENTS

62

OSDs: 10s to 10000s in a cluster One per disk (or one per SSD, RAID group…) Serve stored objects to clients Intelligently peer for replication & recovery

Monitors: Maintain cluster membership and state Provide consensus for distributed decision-

making Small, odd number These do not serve stored objects to clients

M

WHERE DO OBJECTS LIVE?

63

??

APPLICATION

M

M

M

OBJECT

A METADATA SERVER?

64

1

APPLICATION

M

M

M

2

CALCULATED PLACEMENT

65

FAPPLICATION

M

M

MA-G

H-N

O-T

U-Z

EVEN BETTER: CRUSH!

66

RADOS CLUSTER

OBJECT

10

01

01

10

10

01

11

01

10

01

01

10

10

01 11

01

1001

0110 10 01

11

01

CRUSH IS A QUICK CALCULATION

67

RADOS CLUSTER

OBJECT

10

01

01

10

10

01 11

01

1001

0110 10 01

11

01

CRUSH: DYNAMIC DATA PLACEMENT

68

CRUSH: Pseudo-random placement algorithm

Fast calculation, no lookup Repeatable, deterministic

Statistically uniform distribution Stable mapping

Limited data migration on change Rule-based configuration

Infrastructure topology aware Adjustable replication Weighting



69

RGWA web services


with S3 and Swift








APP HOST/VM CLIENT

ACCESSING A RADOS CLUSTER

70

APPLICATION

M M

M

RADOS CLUSTER

LIBRADOS

OBJECT

socket

L

LIBRADOS: RADOS ACCESS FOR APPS

71

LIBRADOS: Direct access to RADOS for applications C, C++, Python, PHP, Java, Erlang Direct access to storage nodes No HTTP overhead



72

RGWA web services


with S3 and Swift








APP HOST/VM CLIENT

THE RADOS GATEWAY

73

M M

M

RADOS CLUSTER

RADOSGW

LIBRADOS

socket

RADOSGW

LIBRADOS

APPLICATION APPLICATION

REST

RADOSGW MAKES RADOS WEBBY

74

RADOSGW: REST-based object storage proxy Uses RADOS to store objects API supports buckets, accounts Usage accounting for billing Compatible with S3 and Swift applications



75

RGWA web services


with S3 and Swift








APP HOST/VM CLIENT

STORING VIRTUAL DISKS

76

M M

RADOS CLUSTER

HYPERVISOR

LIBRBD

VM

SEPARATE COMPUTE FROM STORAGE

77

M M

RADOS CLUSTER

HYPERVISOR

LIBRBDVM

HYPERVISOR

LIBRBD

KERNEL MODULE FOR MAX FLEXIBLE!

78

M M

RADOS CLUSTER

LINUX HOST

KRBD

RBD STORES VIRTUAL DISKS

79

RADOS BLOCK DEVICE: Storage of disk images in RADOS Decouples VMs from host Images are striped across the cluster (pool) Snapshots Copy-on-write clones Support in:

Mainline Linux Kernel (2.6.39+) Qemu/KVM, native Xen coming soon OpenStack, CloudStack, Nebula, Proxmox



80

RGWA web services


with S3 and Swift








APP HOST/VM CLIENT

SEPARATE METADATA SERVER

81

LINUX HOST

M M

M

RADOS CLUSTER

KERNEL MODULE

datametadata 0110

SCALABLE METADATA SERVERS

82

METADATA SERVER Manages metadata for a POSIX-compliant

shared filesystem Directory hierarchy File metadata (owner, timestamps, mode,

etc.) Stores metadata in RADOS Does not serve file data to clients Only required for shared filesystem

CEPH AND OPENSTACK

83

RADOSGWLIBRADOS

M M

RADOS CLUSTER

OPENSTACK

KEYSTONE CINDER GLANCE

NOVASWIFTLIBRB

DLIBRB

D

HYPER- VISOR

LIBRBD

Read about the latest version of Ceph. The latest stuff is always at http://ceph.com/get

Deploy a test cluster using ceph-deploy. Read the quick-start guide at http://ceph.com/qsg

Read the rest of the docs! Find docs for the latest release at http://ceph.com/docs

Ask for help when you get stuck! Community volunteers are waiting for you at

http://ceph.com/help


GETTING STARTED WITH CEPH

84

http://ceph.com/get

http://ceph.com/get

http://ceph.com/qsg

http://ceph.com/qsg

http://ceph.com/qsg

http://ceph.com/qsg

http://ceph.com/docs

http://ceph.com/docs

http://ceph.com/help

Ceph Day London 2014 - The current state of CephFS development

Technology

Transcript of Ceph Day London 2014 - The current state of CephFS development