1Protocol 2-------- 3 4Clients of memcached communicate with server through TCP connections. 5(A UDP interface is also available; details are below under "UDP 6protocol.") A given running memcached server listens on some 7(configurable) port; clients connect to that port, send commands to 8the server, read responses, and eventually close the connection. 9 10There is no need to send any command to end the session. A client may 11just close the connection at any moment it no longer needs it. Note, 12however, that clients are encouraged to cache their connections rather 13than reopen them every time they need to store or retrieve data. This 14is because memcached is especially designed to work very efficiently 15with a very large number (many hundreds, more than a thousand if 16necessary) of open connections. Caching connections will eliminate the 17overhead associated with establishing a TCP connection (the overhead 18of preparing for a new connection on the server side is insignificant 19compared to this). 20 21There are two kinds of data sent in the memcache protocol: text lines 22and unstructured data. Text lines are used for commands from clients 23and responses from servers. Unstructured data is sent when a client 24wants to store or retrieve data. The server will transmit back 25unstructured data in exactly the same way it received it, as a byte 26stream. The server doesn't care about byte order issues in 27unstructured data and isn't aware of them. There are no limitations on 28characters that may appear in unstructured data; however, the reader 29of such data (either a client or a server) will always know, from a 30preceding text line, the exact length of the data block being 31transmitted. 32 33Text lines are always terminated by \r\n. Unstructured data is _also_ 34terminated by \r\n, even though \r, \n or any other 8-bit characters 35may also appear inside the data. Therefore, when a client retrieves 36data from a server, it must use the length of the data block (which it 37will be provided with) to determine where the data block ends, and not 38the fact that \r\n follows the end of the data block, even though it 39does. 40 41Keys 42---- 43 44Data stored by memcached is identified with the help of a key. A key 45is a text string which should uniquely identify the data for clients 46that are interested in storing and retrieving it. Currently the 47length limit of a key is set at 250 characters (of course, normally 48clients wouldn't need to use such long keys); the key must not include 49control characters or whitespace. 50 51Commands 52-------- 53 54There are three types of commands. 55 56Storage commands (there are six: "set", "add", "replace", "append" 57"prepend" and "cas") ask the server to store some data identified by a 58key. The client sends a command line, and then a data block; after 59that the client expects one line of response, which will indicate 60success or failure. 61 62Retrieval commands (there are two: "get" and "gets") ask the server to 63retrieve data corresponding to a set of keys (one or more keys in one 64request). The client sends a command line, which includes all the 65requested keys; after that for each item the server finds it sends to 66the client one response line with information about the item, and one 67data block with the item's data; this continues until the server 68finished with the "END" response line. 69 70All other commands don't involve unstructured data. In all of them, 71the client sends one command line, and expects (depending on the 72command) either one line of response, or several lines of response 73ending with "END" on the last line. 74 75A command line always starts with the name of the command, followed by 76parameters (if any) delimited by whitespace. Command names are 77lower-case and are case-sensitive. 78 79Expiration times 80---------------- 81 82Some commands involve a client sending some kind of expiration time 83(relative to an item or to an operation requested by the client) to 84the server. In all such cases, the actual value sent may either be 85Unix time (number of seconds since January 1, 1970, as a 32-bit 86value), or a number of seconds starting from current time. In the 87latter case, this number of seconds may not exceed 60*60*24*30 (number 88of seconds in 30 days); if the number sent by a client is larger than 89that, the server will consider it to be real Unix time value rather 90than an offset from current time. 91 92 93Error strings 94------------- 95 96Each command sent by a client may be answered with an error string 97from the server. These error strings come in three types: 98 99- "ERROR\r\n" 100 101 means the client sent a nonexistent command name. 102 103- "CLIENT_ERROR <error>\r\n" 104 105 means some sort of client error in the input line, i.e. the input 106 doesn't conform to the protocol in some way. <error> is a 107 human-readable error string. 108 109- "SERVER_ERROR <error>\r\n" 110 111 means some sort of server error prevents the server from carrying 112 out the command. <error> is a human-readable error string. In cases 113 of severe server errors, which make it impossible to continue 114 serving the client (this shouldn't normally happen), the server will 115 close the connection after sending the error line. This is the only 116 case in which the server closes a connection to a client. 117 118 119In the descriptions of individual commands below, these error lines 120are not again specifically mentioned, but clients must allow for their 121possibility. 122 123 124Storage commands 125---------------- 126 127First, the client sends a command line which looks like this: 128 129<command name> <key> <flags> <exptime> <bytes> [noreply]\r\n 130cas <key> <flags> <exptime> <bytes> <cas unique> [noreply]\r\n 131 132- <command name> is "set", "add", "replace", "append" or "prepend" 133 134 "set" means "store this data". 135 136 "add" means "store this data, but only if the server *doesn't* already 137 hold data for this key". 138 139 "replace" means "store this data, but only if the server *does* 140 already hold data for this key". 141 142 "append" means "add this data to an existing key after existing data". 143 144 "prepend" means "add this data to an existing key before existing data". 145 146 The append and prepend commands do not accept flags or exptime. 147 They update existing data portions, and ignore new flag and exptime 148 settings. 149 150 "cas" is a check and set operation which means "store this data but 151 only if no one else has updated since I last fetched it." 152 153- <key> is the key under which the client asks to store the data 154 155- <flags> is an arbitrary 16-bit unsigned integer (written out in 156 decimal) that the server stores along with the data and sends back 157 when the item is retrieved. Clients may use this as a bit field to 158 store data-specific information; this field is opaque to the server. 159 Note that in memcached 1.2.1 and higher, flags may be 32-bits, instead 160 of 16, but you might want to restrict yourself to 16 bits for 161 compatibility with older versions. 162 163- <exptime> is expiration time. If it's 0, the item never expires 164 (although it may be deleted from the cache to make place for other 165 items). If it's non-zero (either Unix time or offset in seconds from 166 current time), it is guaranteed that clients will not be able to 167 retrieve this item after the expiration time arrives (measured by 168 server time). 169 170- <bytes> is the number of bytes in the data block to follow, *not* 171 including the delimiting \r\n. <bytes> may be zero (in which case 172 it's followed by an empty data block). 173 174- <cas unique> is a unique 64-bit value of an existing entry. 175 Clients should use the value returned from the "gets" command 176 when issuing "cas" updates. 177 178- "noreply" optional parameter instructs the server to not send the 179 reply. NOTE: if the request line is malformed, the server can't 180 parse "noreply" option reliably. In this case it may send the error 181 to the client, and not reading it on the client side will break 182 things. Client should construct only valid requests. 183 184After this line, the client sends the data block: 185 186<data block>\r\n 187 188- <data block> is a chunk of arbitrary 8-bit data of length <bytes> 189 from the previous line. 190 191After sending the command line and the data blockm the client awaits 192the reply, which may be: 193 194- "STORED\r\n", to indicate success. 195 196- "NOT_STORED\r\n" to indicate the data was not stored, but not 197because of an error. This normally means that the 198condition for an "add" or a "replace" command wasn't met. 199 200- "EXISTS\r\n" to indicate that the item you are trying to store with 201a "cas" command has been modified since you last fetched it. 202 203- "NOT_FOUND\r\n" to indicate that the item you are trying to store 204with a "cas" command did not exist. 205 206 207Retrieval command: 208------------------ 209 210The retrieval commands "get" and "gets" operates like this: 211 212get <key>*\r\n 213gets <key>*\r\n 214 215- <key>* means one or more key strings separated by whitespace. 216 217After this command, the client expects zero or more items, each of 218which is received as a text line followed by a data block. After all 219the items have been transmitted, the server sends the string 220 221"END\r\n" 222 223to indicate the end of response. 224 225Each item sent by the server looks like this: 226 227VALUE <key> <flags> <bytes> [<cas unique>]\r\n 228<data block>\r\n 229 230- <key> is the key for the item being sent 231 232- <flags> is the flags value set by the storage command 233 234- <bytes> is the length of the data block to follow, *not* including 235 its delimiting \r\n 236 237- <cas unique> is a unique 64-bit integer that uniquely identifies 238 this specific item. 239 240- <data block> is the data for this item. 241 242If some of the keys appearing in a retrieval request are not sent back 243by the server in the item list this means that the server does not 244hold items with such keys (because they were never stored, or stored 245but deleted to make space for more items, or expired, or explicitly 246deleted by a client). 247 248 249Deletion 250-------- 251 252The command "delete" allows for explicit deletion of items: 253 254delete <key> [noreply]\r\n 255 256- <key> is the key of the item the client wishes the server to delete 257 258- "noreply" optional parameter instructs the server to not send the 259 reply. See the note in Storage commands regarding malformed 260 requests. 261 262The response line to this command can be one of: 263 264- "DELETED\r\n" to indicate success 265 266- "NOT_FOUND\r\n" to indicate that the item with this key was not 267 found. 268 269See the "flush_all" command below for immediate invalidation 270of all existing items. 271 272 273Increment/Decrement 274------------------- 275 276Commands "incr" and "decr" are used to change data for some item 277in-place, incrementing or decrementing it. The data for the item is 278treated as decimal representation of a 64-bit unsigned integer. If 279the current data value does not conform to such a representation, the 280incr/decr commands return an error (memcached <= 1.2.6 treated the 281bogus value as if it were 0, leading to confusing). Also, the item 282must already exist for incr/decr to work; these commands won't pretend 283that a non-existent key exists with value 0; instead, they will fail. 284 285The client sends the command line: 286 287incr <key> <value> [noreply]\r\n 288 289or 290 291decr <key> <value> [noreply]\r\n 292 293- <key> is the key of the item the client wishes to change 294 295- <value> is the amount by which the client wants to increase/decrease 296the item. It is a decimal representation of a 64-bit unsigned integer. 297 298- "noreply" optional parameter instructs the server to not send the 299 reply. See the note in Storage commands regarding malformed 300 requests. 301 302The response will be one of: 303 304- "NOT_FOUND\r\n" to indicate the item with this value was not found 305 306- <value>\r\n , where <value> is the new value of the item's data, 307 after the increment/decrement operation was carried out. 308 309Note that underflow in the "decr" command is caught: if a client tries 310to decrease the value below 0, the new value will be 0. Overflow in 311the "incr" command will wrap around the 64 bit mark. 312 313Note also that decrementing a number such that it loses length isn't 314guaranteed to decrement its returned length. The number MAY be 315space-padded at the end, but this is purely an implementation 316optimization, so you also shouldn't rely on that. 317 318Statistics 319---------- 320 321The command "stats" is used to query the server about statistics it 322maintains and other internal data. It has two forms. Without 323arguments: 324 325stats\r\n 326 327it causes the server to output general-purpose statistics and 328settings, documented below. In the other form it has some arguments: 329 330stats <args>\r\n 331 332Depending on <args>, various internal data is sent by the server. The 333kinds of arguments and the data sent are not documented in this vesion 334of the protocol, and are subject to change for the convenience of 335memcache developers. 336 337 338General-purpose statistics 339-------------------------- 340 341Upon receiving the "stats" command without arguments, the server sents 342a number of lines which look like this: 343 344STAT <name> <value>\r\n 345 346The server terminates this list with the line 347 348END\r\n 349 350In each line of statistics, <name> is the name of this statistic, and 351<value> is the data. The following is the list of all names sent in 352response to the "stats" command, together with the type of the value 353sent for this name, and the meaning of the value. 354 355In the type column below, "32u" means a 32-bit unsigned integer, "64u" 356means a 64-bit unsigner integer. '32u.32u' means two 32-but unsigned 357integers separated by a colon (treat this as a floating point number). 358 359|-----------------------+---------+-------------------------------------------| 360| Name | Type | Meaning | 361|-----------------------+---------+-------------------------------------------| 362| pid | 32u | Process id of this server process | 363| uptime | 32u | Number of secs since the server started | 364| time | 32u | current UNIX time according to the server | 365| version | string | Version string of this server | 366| pointer_size | 32 | Default size of pointers on the host OS | 367| | | (generally 32 or 64) | 368| rusage_user | 32u.32u | Accumulated user time for this process | 369| | | (seconds:microseconds) | 370| rusage_system | 32u.32u | Accumulated system time for this process | 371| | | (seconds:microseconds) | 372| curr_items | 32u | Current number of items stored | 373| total_items | 32u | Total number of items stored since | 374| | | the server started | 375| bytes | 64u | Current number of bytes used | 376| | | to store items | 377| curr_connections | 32u | Number of open connections | 378| total_connections | 32u | Total number of connections opened since | 379| | | the server started running | 380| connection_structures | 32u | Number of connection structures allocated | 381| | | by the server | 382| cmd_get | 64u | Cumulative number of retrieval reqs | 383| cmd_set | 64u | Cumulative number of storage reqs | 384| get_hits | 64u | Number of keys that have been requested | 385| | | and found present | 386| get_misses | 64u | Number of items that have been requested | 387| | | and not found | 388| delete_misses | 64u | Number of deletions reqs for missing keys | 389| delete_hits | 64u | Number of deletion reqs resulting in | 390| | | an item being removed. | 391| incr_misses | 64u | Number of incr reqs against missing keys. | 392| incr_hits | 64u | Number of successful incr reqs. | 393| decr_misses | 64u | Number of decr reqs against missing keys. | 394| decr_hits | 64u | Number of successful decr reqs. | 395| cas_misses | 64u | Number of CAS reqs against missing keys. | 396| cas_hits | 64u | Number of successful CAS reqs. | 397| cas_badval | 64u | Number of CAS reqs for which a key was | 398| | | found, but the CAS value did not match. | 399| evictions | 64u | Number of valid items removed from cache | 400| | | to free memory for new items | 401| bytes_read | 64u | Total number of bytes read by this server | 402| | | from network | 403| bytes_written | 64u | Total number of bytes sent by this server | 404| | | to network | 405| limit_maxbytes | 32u | Number of bytes this server is allowed to | 406| | | use for storage. | 407| threads | 32u | Number of worker threads requested. | 408| | | (see doc/threads.txt) | 409| conn_yields | 64u | Number of times any connection yielded to | 410| | | another due to hitting the -R limit. | 411|-----------------------+---------+-------------------------------------------| 412 413Settings statistics 414------------------- 415CAVEAT: This section describes statistics which are subject to change in the 416future. 417 418The "stats" command with the argument of "settings" returns details of 419the settings of the running memcached. This is primarily made up of 420the results of processing commandline options. 421 422Note that these are not guaranteed to return in any specific order and 423this list may not be exhaustive. Otherwise, this returns like any 424other stats command. 425 426|-----------------+----------+----------------------------------------------| 427| Name | Type | Meaning | 428|-----------------+----------+----------------------------------------------| 429| maxbytes | size_t | Maximum number of bytes allows in this cache | 430| maxconns | 32 | Maximum number of clients allowed. | 431| tcpport | 32 | TCP listen port. | 432| udpport | 32 | UDP listen port. | 433| inter | string | Listen interface. | 434| verbosity | 32 | 0 = none, 1 = some, 2 = lots | 435| oldest | 32u | Age of the oldest honored object. | 436| evictions | on/off | When off, LRU evictions are disabled. | 437| domain_socket | string | Path to the domain socket (if any). | 438| umask | 32 (oct) | umask for the creation of the domain socket. | 439| growth_factor | float | Chunk size growth factor. | 440| chunk_size | 32 | Minimum space allocated for key+value+flags | 441| num_threads | 32 | Number of threads (including dispatch). | 442| stat_key_prefix | char | Stats prefix separator character. | 443| detail_enabled | bool | If yes, stats detail is enabled. | 444| reqs_per_event | 32 | Max num IO ops processed within an event. | 445| cas_enabled | bool | When no, CAS is not enabled for this server. | 446| tcp_backlog | 32 | TCP listen backlog. 447|-----------------+----------+----------------------------------------------| 448 449 450Item statistics 451--------------- 452CAVEAT: This section describes statistics which are subject to change in the 453future. 454 455The "stats" command with the argument of "items" returns information about 456item storage per slab class. The data is returned in the format: 457 458STAT items:<slabclass>:<stat> <value>\r\n 459 460The server terminates this list with the line 461 462END\r\n 463 464The slabclass aligns with class ids used by the "stats slabs" command. Where 465"stats slabs" describes size and memory usage, "stats items" shows higher 466level information. 467 468The following item values are defined as of writing. 469 470Name Meaning 471------------------------------ 472number Number of items presently stored in this class. Expired 473 items are not automatically excluded. 474age Age of the oldest item in the LRU. 475evicted Number of times an item had to be evicted from the LRU 476 before it expired. 477evicted_nonzero Number of times an item which had an explicit expire 478 time set had to be evicted from the LRU before it 479 expired. 480evicted_time Seconds since the last access for the most recent item 481 evicted from this class. Use this to judge how 482 recently active your evicted data is. 483outofmemory Number of times the underlying slab class was unable to 484 store a new item. This means you are running with -M or 485 an eviction failed. 486tailrepairs Number of times we self-healed a slab with a refcount 487 leak. If this counter is increasing a lot, please 488 report your situation to the developers. 489 490Note this will only display information about slabs which exist, so an empty 491cache will return an empty set. 492 493 494Item size statistics 495-------------------- 496CAVEAT: This section describes statistics which are subject to change in the 497future. 498 499The "stats" command with the argument of "sizes" returns information about the 500general size and count of all items stored in the cache. 501WARNING: This command WILL lock up your cache! It iterates over *every item* 502and examines the size. While the operation is fast, if you have many items 503you could prevent memcached from serving requests for several seconds. 504 505The data is returned in the following format: 506 507<size> <count>\r\n 508 509The server terminates this list with the line 510 511END\r\n 512 513'size' is an approximate size of the item, within 32 bytes. 514'count' is the amount of items that exist within that 32-byte range. 515 516This is essentially a display of all of your items if there was a slab class 517for every 32 bytes. You can use this to determine if adjusting the slab growth 518factor would save memory overhead. For example: generating more classes in the 519lower range could allow items to fit more snugly into their slab classes, if 520most of your items are less than 200 bytes in size. 521 522 523Slab statistics 524--------------- 525CAVEAT: This section describes statistics which are subject to change in the 526future. 527 528The "stats" command with the argument of "slabs" returns information about 529each of the slabs created by memcached during runtime. This includes per-slab 530information along with some totals. The data is returned in the format: 531 532STAT <slabclass>:<stat> <value>\r\n 533STAT <stat> <value>\r\n 534 535The server terminates this list with the line 536 537END\r\n 538 539|-----------------+----------------------------------------------------------| 540| Name | Meaning | 541|-----------------+----------------------------------------------------------| 542| chunk_size | The amount of space each chunk uses. One item will use | 543| | one chunk of the appropriate size. | 544| chunks_per_page | How many chunks exist within one page. A page by | 545| | default is one megabyte in size. Slabs are allocated per | 546| | page, then broken into chunks. | 547| total_pages | Total number of pages allocated to the slab class. | 548| total_chunks | Total number of chunks allocated to the slab class. | 549| get_hits | Total number of get requests serviced by this class. | 550| cmd_set | Total number of set requests storing data in this class. | 551| delete_hits | Total number of successful deletes from this class. | 552| incr_hits | Total number of incrs modifying this class. | 553| decr_hits | Total number of decrs modifying this class. | 554| cas_hits | Total number of CAS commands modifying this class. | 555| cas_badval | Total number of CAS commands that failed to modify a | 556| | value due to a bad CAS id. | 557| used_chunks | How many chunks have been allocated to items. | 558| free_chunks | Chunks not yet allocated to items, or freed via delete. | 559| free_chunks_end | Number of free chunks at the end of the last allocated | 560| | page. | 561| mem_requested | Number of bytes requested to be stored in this slab[*]. | 562| active_slabs | Total number of slab classes allocated. | 563| total_malloced | Total amount of memory allocated to slab pages. | 564|-----------------+----------------------------------------------------------| 565 566* Items are stored in a slab that is the same size or larger than the 567 item. mem_requested shows the size of all items within a 568 slab. (total_pages * 1MB) - mem_requested shows memory wasted in a 569 slab class. If you see a lot of waste, consider tuning the slab 570 factor. 571 572Other commands 573-------------- 574 575"flush_all" is a command with an optional numeric argument. It always 576succeeds, and the server sends "OK\r\n" in response (unless "noreply" 577is given as the last parameter). Its effect is to invalidate all 578existing items immediately (by default) or after the expiration 579specified. After invalidation none of the items will be returned in 580response to a retrieval command (unless it's stored again under the 581same key *after* flush_all has invalidated the items). flush_all 582doesn't actually free all the memory taken up by existing items; that 583will happen gradually as new items are stored. The most precise 584definition of what flush_all does is the following: it causes all 585items whose update time is earlier than the time at which flush_all 586was set to be executed to be ignored for retrieval purposes. 587 588The intent of flush_all with a delay, was that in a setting where you 589have a pool of memcached servers, and you need to flush all content, 590you have the option of not resetting all memcached servers at the 591same time (which could e.g. cause a spike in database load with all 592clients suddenly needing to recreate content that would otherwise 593have been found in the memcached daemon). 594 595The delay option allows you to have them reset in e.g. 10 second 596intervals (by passing 0 to the first, 10 to the second, 20 to the 597third, etc. etc.). 598 599 600"version" is a command with no arguments: 601 602version\r\n 603 604In response, the server sends 605 606"VERSION <version>\r\n", where <version> is the version string for the 607server. 608 609"verbosity" is a command with a numeric argument. It always succeeds, 610and the server sends "OK\r\n" in response (unless "noreply" is given 611as the last parameter). Its effect is to set the verbosity level of 612the logging output. 613 614"quit" is a command with no arguments: 615 616quit\r\n 617 618Upon receiving this command, the server closes the 619connection. However, the client may also simply close the connection 620when it no longer needs it, without issuing this command. 621 622 623UDP protocol 624------------ 625 626For very large installations where the number of clients is high enough 627that the number of TCP connections causes scaling difficulties, there is 628also a UDP-based interface. The UDP interface does not provide guaranteed 629delivery, so should only be used for operations that aren't required to 630succeed; typically it is used for "get" requests where a missing or 631incomplete response can simply be treated as a cache miss. 632 633Each UDP datagram contains a simple frame header, followed by data in the 634same format as the TCP protocol described above. In the current 635implementation, requests must be contained in a single UDP datagram, but 636responses may span several datagrams. (The only common requests that would 637span multiple datagrams are huge multi-key "get" requests and "set" 638requests, both of which are more suitable to TCP transport for reliability 639reasons anyway.) 640 641The frame header is 8 bytes long, as follows (all values are 16-bit integers 642in network byte order, high byte first): 643 6440-1 Request ID 6452-3 Sequence number 6464-5 Total number of datagrams in this message 6476-7 Reserved for future use; must be 0 648 649The request ID is supplied by the client. Typically it will be a 650monotonically increasing value starting from a random seed, but the client 651is free to use whatever request IDs it likes. The server's response will 652contain the same ID as the incoming request. The client uses the request ID 653to differentiate between responses to outstanding requests if there are 654several pending from the same server; any datagrams with an unknown request 655ID are probably delayed responses to an earlier request and should be 656discarded. 657 658The sequence number ranges from 0 to n-1, where n is the total number of 659datagrams in the message. The client should concatenate the payloads of the 660datagrams for a given response in sequence number order; the resulting byte 661stream will contain a complete response in the same format as the TCP 662protocol (including terminating \r\n sequences). 663