1MAILDIR(5) Double Precision, Inc. MAILDIR(5)
2
6 maildir - E-mail directory
7
9 $HOME/Maildir
10
12 A “Maildir” is a structured directory that holds E-mail messages.
13 Maildirs were first implemented by the Qmail mail server. Qmail´s
14 maildirs were a simple data structure, nothing more than a single
15 collection of E-mail messages. The Courier mail server builds upon
16 Qmail´s maildirs to provide extended functionality, such as folders and
17 quotas. This document describes the Courier mail server´s extended
18 maildirs, without explicitly identifying The Courier mail
19 server-specific extensions. See maildir(5) in Qmail´s documentation for
20 the original definition of maildirs.
21 Traditionally, E-mail folders were saved as plain text files, called
23 “mboxes”. Mboxes have known limitations. Only one application can use
24 an mbox at the same time. Locking is required in order to allow
25 simultaneous concurrent access by different applications. Locking is
26 often problematic, and not very reliable in network-based filesystem
27 requirements. Some network-based filesystems don´t offer any reliable
28 locking mechanism at all. Furthermore, even bulletproof locking won´t
29 prevent occasional mbox corruption. A process can be killed or
30 terminated in the middle of updating an mbox. This will likely result
31 in corruption, and a loss of most messages in the mbox.
32 Maildirs allow multiple concurrent access by different applications.
34 Maildirs do not require locking. Multiple applications can update a
35 maildir at the same time, without stepping on each other´s feet.
36 Maildir contents
38 A “maildir” is a directory that´s created by maildirmake(1)[1].
39 Naturally, maildirs should not have any group or world permissions,
40 unless you want other people to read your mail. A maildir contains
41 three subdirectories: tmp, new, and cur. These three subdirectories
42 comprise the primary folder, where new mail is delivered by the system.
43 Folders are additional subdirectories in the maildir whose names begin
45 with a period: such as .Drafts or .Sent. Each folder itself contains
46 the same three subdirectories, tmp, new, and cur, and an additional
47 zero-length file named maildirfolder, whose purpose is to inform any
48 mail delivery agent that it´s really delivering to a folder, and that
49 the mail delivery agent should look in the parent directory for any
50 maildir-related information.
51 Folders are not physically nested. A folder subdirectory, such as .Sent
53 does not itself contain any subfolders. The main maildir contains a
54 single, flat list of subfolders. These folders are logically nested,
55 and periods serve to separate folder hierarchies. For example,
56 .Sent.2002 is considered to be a subfolder called “2002” which is a
57 subfolder of “Sent”.
58 Folder name encoding
60 Folder names can contain any Unicode character, except for control
61 characters. US-ASCII characters, U+0x0020 - U+0x007F, except for
62 the period, forward-slash, and ampersand characters (U+0x002E,
63 U+0x002F, and U+0x0026) represent themselves. The ampersand is
64 represent by the two character sequence “&-”. The period, forward
65 slash, and non US-ASCII Unicode characters are represented using
66 the UTF-7 character set, and encoded with a modified form of
67 base64-encoding.
68 The “&” character starts the modified base64-encoded sequence; the
70 sequence is terminated by the “-” character. The sequence of 16-bit
71 Unicode characters is written in big-endian order, and encoded
72 using the base64-encoding method described in section 5.2 of RFC
73 1521[2], with the following modifications:
74 · The “=” padding character is omitted. When decoding, an
76 incomplete 16-bit character is discarded.
77 · The comma character, “,” is used in place of the “/” character
79 in the base64 alphabet.
80 For example, the word “Resume” with both "e"s being the e-acute
82 character, U+0x00e9, is encoded as “R&AOk-sum&AOk-” (so a folder of
83 that name would be a maildir subdirectory called
84 “.R&AOk-sum&AOk-”).
85 Other maildir contents
87 Software that uses maildirs may also create additional files
88 besides the tmp, new, and cur subdirectories -- in the main maildir
89 or a subfolder -- for its own purposes.
90 Messages
92 E-mail messages are stored in separate, individual files, one E-mail
93 message per file. The tmp subdirectory temporarily stores E-mail
94 messages that are in the process of being delivered to this maildir.
95 tmp may also store other kinds of temporary files, as long as they are
96 created in the same way that message files are created in tmp. The new
97 subdirectory stores messages that have been delivered to this maildir,
98 but have not yet been seen by any mail application. The cur
99 subdirectory stores messages that have already been seen by mail
100 applications.
101 Adding new mail to maildirs
103 The following process delivers a new message to the maildir:
104 A new unique filename is created using one of two possible forms:
106 “time.MusecPpid.host”, or “time.MusecPpid_unique.host”. “time” and
107 “usec” is the current system time, obtained from gettimeofday(2).
108 “pid” is the process number of the process that is delivering this
109 message to the maildir. “host” is the name of the machine where the
110 mail is being delivered. In the event that the same process creates
111 multiple messages, a suffix unique to each message is appended to the
112 process id; preferrably an underscore, followed by an increasing
113 counter. This applies whether messages created by a process are all
114 added to the same, or different, maildirs. This protocol allows
115 multiple processes running on multiple machines on the same network to
116 simultaneously create new messages without stomping on each other.
117 The filename created in the previous step is checked for existence by
119 executing the stat(2) system call. If stat(2) results in ANYTHING OTHER
120 than the system error ENOENT, the process must sleep for two seconds,
121 then go back and create another unique filename. This is an extra step
122 to insure that each new message has a completely unique filename.
123 Other applications that wish to use tmp for temporary storage should
125 observe the same protocol (but see READING MAIL FROM MAILDIRS below,
126 because old files in tmp will be eventually deleted).
127 If the stat(2) system call returned ENOENT, the process may proceed to
129 create the file in the tmp subdirectory, and save the entire message in
130 the new file. The message saved MUST NOT have the “From_” header that
131 is used to mboxes. The message also MUST NOT have any “From_” lines in
132 the contents of the message prefixed by the “>” character.
133 When saving the message, the number of bytes returned by the write(2)
135 system call must be checked, in order to make sure that the complete
136 message has been written out.
137 After the message is saved, the file descriptor is fstat(2)-ed. The
139 file´s device number, inode number, and the its byte size, are saved.
140 The file is closed and is then immediately moved/renamed into the new
141 subdirectory. The name of the file in new should be
142 “time.MusecPpidVdevIino.host,S=cnt”, or
143 “time.MusecPpidVdevIino_unique.host,S=cnt”. “dev” is the message´s
144 device number, “ino” is the message´s inode number (from the previous
145 fstat(2) call); and “cnt” is the message´s size, in bytes.
146 The “,S=cnt” part optimizes the Courier[3] mail server´s maildir quota
148 enhancement; it allows the size of all the mail stored in the maildir
149 to be added up without issuing the stat(2) system call for each
150 individual message (this can be quite a performance drain with certain
151 network filesystems).
152 READING MAIL FROM MAILDIRS
154 Applications that read mail from maildirs should do it in the following
155 order:
156 When opening a maildir or a maildir folder, read the tmp subdirectory
158 and delete any files in there that are at least 36 hours old.
159 Look for new messages in the new subdirectory. Rename new/filename, as
161 cur/filename:2,info. Here, info represents the state of the message,
162 and it consists of zero or more boolean flags chosen from the
163 following: “D” - this is a ´draft´ message, “R” - this message has been
164 replied to, “S” - this message has been viewed (seen), “T” - this
165 message has been marked to be deleted (trashed), but is not yet removed
166 (messages are removed from maildirs simply by deleting their file), “F”
167 - this message has been marked by the user, for some purpose. These
168 flags must be stored in alphabetical order. New messages contain only
169 the :2, suffix, with no flags, indicating that the messages were not
170 seen, replied, marked, or deleted.
171 Maildirs may have maximum size quotas defined, but these quotas are
173 purely voluntary. If you need to implement mandatory quotas, you should
174 use any quota facilities provided by the underlying filesystem that is
175 used to store the maildirs. The maildir quota enhancement is designed
176 to be used in certain situations where filesystem-based quotas cannot
177 be used for some reason. The implementation is designed to avoid the
178 use of any locking. As such, at certain times the calculated quota may
179 be imprecise, and certain anomalous situations may result in the
180 maildir actually going over the stated quota. One such situation would
181 be when applications create messages without updating the quota
182 estimate for the maildir. Eventually it will be precisely recalculated,
183 but wherever possible new messages should be created in compliance with
184 the voluntary quota protocol.
185 The voluntary quota protocol involves some additional procedures that
187 must be followed when creating or deleting messages within a given
188 maildir or its subfolders. The deliverquota(8)[4] command is a tiny
189 application that delivers a single message to a maildir using the
190 voluntary quota protocol, and hopefully it can be used as a measure of
191 last resort. Alternatively, applications can use the libmaildir.a
192 library to handle all the low-level dirty details for them. The
193 voluntary quota enhancement is described in the maildirquota(7)[5] man
194 page.
195 Maildir Quotas
197 This is a voluntary mechanism for enforcing "loose" quotas on the
198 maximum sizes of maildirs. This mechanism is enforced in software, and
199 not by the operating system. Therefore it is only effective as long as
200 the maildirs themselves are not directly accessible by their users,
201 since this mechanism is trivially disabled.
202 If possible, operating system-enforced quotas are preferrable. Where
204 operating system quota enforcement is not available, or not possible,
205 this voluntary quota enforcement mechanism might be an acceptable
206 compromise. Since it´s enforced in software, all software that modifies
207 or accesses the maildirs is required to voluntary obey and enforce a
208 quota. The voluntary quota implementation is flexible enough to allow
209 non quota-aware applications to also access the maildirs, without any
210 drastic consequences. There will be some non-drastic consequences,
211 though. Of course, non quota-aware applications will not enforce any
212 defined quotas. Furthermore, this voluntary maildir quota mechanism
213 works by estimating the current size of the maildir, with periodic
214 exact recalculation. Obviously non quota-aware maildir applications
215 will not update the maildir size estimation, so the estimate will be
216 thrown off for some period of time, until the next recalculation.
217 This voluntary quota mechanism is designed to be a reasonable
219 compromise between effectiveness, and performance. The entire purpose
220 of using maildir-based mail storage is to avoid any kind of locking,
221 and to permit parallel access to mail by multiple applications. In
222 order to compute the exact size of a maildir, the maildir must be
223 locked somehow to prevent any modifications while its contents are
224 added up. Obviously something like that defeats the original purpose of
225 using maildirs, therefore the voluntary quota mechanism does not use
226 locking, and that´s why the current recorded maildir size is always
227 considered to be an estimate. Regular size recalculations will
228 compensate for any occasional race conditions that result in the
229 estimate to be thrown off.
230 A quota for an existing maildir is installed by running maildirmake
232 with the -q option, and naming an existing maildir. The -q option takes
233 a parameter, quota, which is a comma-separated list of quota
234 specifications. A quota specification consists of a number followed by
235 either ´S´, indicating the maximum message size in bytes, or ´C´,
236 maximum number of messages. For example:
237 This sets the quota to 5,000,000 bytes or 1000 messages, whichever
239 comes first.
240 This sets the quota to 1,000,000 bytes, without limiting the number of
242 messages.
243 A quota of an existing maildir can be changed by rerunning the
245 maildirmake command with a new -q option. To delete a quota entirely,
246 delete the Maildir/maildirsize file.
247
249 maildirmake(1)[1].
250
252 1. maildirmake(1)
253 [set $man.base.url.for.relative.links]/maildirmake.html
254 2. RFC 1521
256 http://www.rfc-editor.org/rfc/rfc1521.txt
257 3. Courier
259 http://www.courier-mta.org
260 4. deliverquota(8)
262 [set $man.base.url.for.relative.links]/deliverquota.html
263 5. maildirquota(7)
265 [set $man.base.url.for.relative.links]/maildirquota.html
266Double Precision, Inc. 02/19/2010 MAILDIR(5)