222 lines · plain
1.. SPDX-License-Identifier: GPL-2.02 3============================4Ceph Distributed File System5============================6 7Ceph is a distributed network file system designed to provide good8performance, reliability, and scalability.9 10Basic features include:11 12 * POSIX semantics13 * Seamless scaling from 1 to many thousands of nodes14 * High availability and reliability. No single point of failure.15 * N-way replication of data across storage nodes16 * Fast recovery from node failures17 * Automatic rebalancing of data on node addition/removal18 * Easy deployment: most FS components are userspace daemons19 20Also,21 22 * Flexible snapshots (on any directory)23 * Recursive accounting (nested files, directories, bytes)24 25In contrast to cluster filesystems like GFS, OCFS2, and GPFS that rely26on symmetric access by all clients to shared block devices, Ceph27separates data and metadata management into independent server28clusters, similar to Lustre. Unlike Lustre, however, metadata and29storage nodes run entirely as user space daemons. File data is striped30across storage nodes in large chunks to distribute workload and31facilitate high throughputs. When storage nodes fail, data is32re-replicated in a distributed fashion by the storage nodes themselves33(with some minimal coordination from a cluster monitor), making the34system extremely efficient and scalable.35 36Metadata servers effectively form a large, consistent, distributed37in-memory cache above the file namespace that is extremely scalable,38dynamically redistributes metadata in response to workload changes,39and can tolerate arbitrary (well, non-Byzantine) node failures. The40metadata server takes a somewhat unconventional approach to metadata41storage to significantly improve performance for common workloads. In42particular, inodes with only a single link are embedded in43directories, allowing entire directories of dentries and inodes to be44loaded into its cache with a single I/O operation. The contents of45extremely large directories can be fragmented and managed by46independent metadata servers, allowing scalable concurrent access.47 48The system offers automatic data rebalancing/migration when scaling49from a small cluster of just a few nodes to many hundreds, without50requiring an administrator carve the data set into static volumes or51go through the tedious process of migrating data between servers.52When the file system approaches full, new nodes can be easily added53and things will "just work."54 55Ceph includes flexible snapshot mechanism that allows a user to create56a snapshot on any subdirectory (and its nested contents) in the57system. Snapshot creation and deletion are as simple as 'mkdir58.snap/foo' and 'rmdir .snap/foo'.59 60Snapshot names have two limitations:61 62* They can not start with an underscore ('_'), as these names are reserved63 for internal usage by the MDS.64* They can not exceed 240 characters in size. This is because the MDS makes65 use of long snapshot names internally, which follow the format:66 `_<SNAPSHOT-NAME>_<INODE-NUMBER>`. Since filenames in general can't have67 more than 255 characters, and `<node-id>` takes 13 characters, the long68 snapshot names can take as much as 255 - 1 - 1 - 13 = 240.69 70Ceph also provides some recursive accounting on directories for nested files71and bytes. You can run the commands::72 73 getfattr -n ceph.dir.rfiles /some/dir74 getfattr -n ceph.dir.rbytes /some/dir75 76to get the total number of nested files and their combined size, respectively.77This makes the identification of large disk space consumers relatively quick,78as no 'du' or similar recursive scan of the file system is required.79 80Finally, Ceph also allows quotas to be set on any directory in the system.81The quota can restrict the number of bytes or the number of files stored82beneath that point in the directory hierarchy. Quotas can be set using83extended attributes 'ceph.quota.max_files' and 'ceph.quota.max_bytes', eg::84 85 setfattr -n ceph.quota.max_bytes -v 100000000 /some/dir86 getfattr -n ceph.quota.max_bytes /some/dir87 88A limitation of the current quotas implementation is that it relies on the89cooperation of the client mounting the file system to stop writers when a90limit is reached. A modified or adversarial client cannot be prevented91from writing as much data as it needs.92 93Mount Syntax94============95 96The basic mount syntax is::97 98 # mount -t ceph user@fsid.fs_name=/[subdir] mnt -o mon_addr=monip1[:port][/monip2[:port]]99 100You only need to specify a single monitor, as the client will get the101full list when it connects. (However, if the monitor you specify102happens to be down, the mount won't succeed.) The port can be left103off if the monitor is using the default. So if the monitor is at1041.2.3.4::105 106 # mount -t ceph cephuser@07fe3187-00d9-42a3-814b-72a4d5e7d5be.cephfs=/ /mnt/ceph -o mon_addr=1.2.3.4107 108is sufficient. If /sbin/mount.ceph is installed, a hostname can be109used instead of an IP address and the cluster FSID can be left out110(as the mount helper will fill it in by reading the ceph configuration111file)::112 113 # mount -t ceph cephuser@cephfs=/ /mnt/ceph -o mon_addr=mon-addr114 115Multiple monitor addresses can be passed by separating each address with a slash (`/`)::116 117 # mount -t ceph cephuser@cephfs=/ /mnt/ceph -o mon_addr=192.168.1.100/192.168.1.101118 119When using the mount helper, monitor address can be read from ceph120configuration file if available. Note that, the cluster FSID (passed as part121of the device string) is validated by checking it with the FSID reported by122the monitor.123 124Mount Options125=============126 127 mon_addr=ip_address[:port][/ip_address[:port]]128 Monitor address to the cluster. This is used to bootstrap the129 connection to the cluster. Once connection is established, the130 monitor addresses in the monitor map are followed.131 132 fsid=cluster-id133 FSID of the cluster (from `ceph fsid` command).134 135 ip=A.B.C.D[:N]136 Specify the IP and/or port the client should bind to locally.137 There is normally not much reason to do this. If the IP is not138 specified, the client's IP address is determined by looking at the139 address its connection to the monitor originates from.140 141 wsize=X142 Specify the maximum write size in bytes. Default: 64 MB.143 144 rsize=X145 Specify the maximum read size in bytes. Default: 64 MB.146 147 rasize=X148 Specify the maximum readahead size in bytes. Default: 8 MB.149 150 mount_timeout=X151 Specify the timeout value for mount (in seconds), in the case152 of a non-responsive Ceph file system. The default is 60153 seconds.154 155 caps_max=X156 Specify the maximum number of caps to hold. Unused caps are released157 when number of caps exceeds the limit. The default is 0 (no limit)158 159 rbytes160 When stat() is called on a directory, set st_size to 'rbytes',161 the summation of file sizes over all files nested beneath that162 directory. This is the default.163 164 norbytes165 When stat() is called on a directory, set st_size to the166 number of entries in that directory.167 168 nocrc169 Disable CRC32C calculation for data writes. If set, the storage node170 must rely on TCP's error correction to detect data corruption171 in the data payload.172 173 dcache174 Use the dcache contents to perform negative lookups and175 readdir when the client has the entire directory contents in176 its cache. (This does not change correctness; the client uses177 cached metadata only when a lease or capability ensures it is178 valid.)179 180 nodcache181 Do not use the dcache as above. This avoids a significant amount of182 complex code, sacrificing performance without affecting correctness,183 and is useful for tracking down bugs.184 185 noasyncreaddir186 Do not use the dcache as above for readdir.187 188 noquotadf189 Report overall filesystem usage in statfs instead of using the root190 directory quota.191 192 nocopyfrom193 Don't use the RADOS 'copy-from' operation to perform remote object194 copies. Currently, it's only used in copy_file_range, which will revert195 to the default VFS implementation if this option is used.196 197 recover_session=<no|clean>198 Set auto reconnect mode in the case where the client is blocklisted. The199 available modes are "no" and "clean". The default is "no".200 201 * no: never attempt to reconnect when client detects that it has been202 blocklisted. Operations will generally fail after being blocklisted.203 204 * clean: client reconnects to the ceph cluster automatically when it205 detects that it has been blocklisted. During reconnect, client drops206 dirty data/metadata, invalidates page caches and writable file handles.207 After reconnect, file locks become stale because the MDS loses track208 of them. If an inode contains any stale file locks, read/write on the209 inode is not allowed until applications release all stale file locks.210 211More Information212================213 214For more information on Ceph, see the home page at215 https://ceph.com/216 217The Linux kernel client source tree is available at218 - https://github.com/ceph/ceph-client.git219 220and the source for the full system is at221 https://github.com/ceph/ceph.git222