Mercurial > dovecot > core-2.2
annotate TODO @ 13470:6a918d40d6f7
TODO updated
author | Timo Sirainen <tss@iki.fi> |
---|---|
date | Fri, 16 Sep 2011 16:14:14 +0300 |
parents | f0fb3cede05c |
children | 825a122eb9d3 |
rev | line source |
---|---|
13470 | 1 - mdbox/sdbox index rebuild -> quota rebuild? |
13360 | 2 - anvil crash -> Error: Anvil client not compatible with this server (mixed old and new binaries?) |
3 - solr separate attachments (patch) | |
4 - sql connection pooling: Count lookup latencies, avoid servers with | |
5 significantly higher latencies. | |
6 - maildir_storage_sync_force() shouldn't do anything except find the new | |
7 file, don't go expunging any more stuff or we could get recursively back to | |
8 where we started, and stuff would break | |
9 - imap: append/copy could trigger indexing (optionally) | |
10 - fuzzy: be fuzzy about date/size | |
12970 | 11 - fix proxy_maybe=fqdn |
12756 | 12 - mailbox list index: |
13360 | 13 - indexing name="" probably crashes in index_mailbox_list_sync_name() |
12756 | 14 - with in-memory indexes be sure to refresh it more often |
15 - refreshing could refresh only the parts that are actually requested, | |
16 e.g. % | |
17 - mailbox_get_metadata(guid) could be optimized | |
18 - virtual could use it to avoid keeping all mailboxes open | |
12970 | 19 - imapc: |
13470 | 20 - lda/lmtp crashes because saved mail's seq=0 |
12970 | 21 - prefetching to THREAD and SORT |
22 - after LIST "" * is refreshed, delete all unlisted local index dirs | |
12586
a2780b694b2d
lib-storage: mailbox_alloc() now takes a virtual mailbox name and other related API changes.
Timo Sirainen <tss@iki.fi>
parents:
12579
diff
changeset
|
23 - check: |
a2780b694b2d
lib-storage: mailbox_alloc() now takes a virtual mailbox name and other related API changes.
Timo Sirainen <tss@iki.fi>
parents:
12579
diff
changeset
|
24 - dsyncing between two namespace separators is probably broken.. |
12579 | 25 - remove mail_deliver_session after all, do all the stuff transparently |
26 by hooking into mailbox_copy(). | |
27 - use this hook also to do the mail deduplication: 1) sort all destination | |
28 users, 2) create mail_user only once for each user, 3) remember in | |
29 src_mail the previously copied mail, 4) use that for mailbox_copy()ing | |
30 to following recipients | |
12551 | 31 - notify_sync() could have "what changed" struct with old/new flags |
12546 | 32 - maildir: copy dovecot-shared file from parent mailbox, not root. |
33 - if global ACL path points to a file instead of a directory, read all of | |
34 them from there. support wildcards in names. | |
35 | |
12476 | 36 - master passdb preserves userdb_* extra fields. should it preserve |
37 non-userdb_* extra fields too? | |
12400 | 38 - imap, pop3: if client init fails, wait a second or two before disconnecting |
39 client. | |
40 - doveadm expunge parameter to delete empty mailboxes (for lazy-expunge) | |
41 - doveadm search savedbefore 7d could be optimized in large mailboxes.. | |
12336 | 42 - mdbox: storage rebuilding could log about changes it does |
12400 | 43 - mdbox: broken extrefs header keeps causing index rebuilds |
12336 | 44 - cache: mail_cache_lookup() should be able to return mail_cache_add()ed |
45 fields even before they've been flushed to disk. this is useful when copying | |
46 messages and some plugin (e.g. mail_log) fetches some fields that are | |
47 already added to cache (to avoid opening and parsing the message) | |
12231 | 48 - doveadm -A <<EOF expunge + purge + EOF |
49 - sent, drafts: .Sent/dovecot.index: modseq_hdr.log_offset too large | |
12181 | 50 - mail_max_lock_timeout error could be reported more nicely, also ones coming |
51 from lib-index | |
52 - dsync: if mailbox is renamed due to it being invalid, its subscription | |
53 isn't renamed | |
12095 | 54 - sql pool: if async query is pending and sync query is sent and there |
55 are no more empty connections, it should flush the async query first | |
11683 | 56 - NTLMv1 and LM should be disabled if disable_plaintext_auth=yes |
11387 | 57 - SEARCH SENT*/HEADER/etc. doesn't seem optimized when using with TEXT/BODY |
12476 | 58 - dict sql: support ignoring some search key hierarchies (e.g. acl "anyone") |
11273 | 59 |
11683 | 60 - fts-solr: crashes if expunge is done while search is indexing |
10968 | 61 - fts-solr: handle DELETE, RENAME. use mailbox GUIDs (optionally) |
10546 | 62 |
10479 | 63 - mdbox |
11485 | 64 - dotlocking: cleanup should delete stale *.lock files |
10968 | 65 - purging seems to be inefficient. run imaptest for a while, get >500 |
66 files, start purging, it's slow until there are about 100 files left, | |
67 then the rest is suddenly fast. | |
68 - make sure that when reading mdbox mails sequentially the data is being | |
69 read from disk in n kB blocks and reads cross mail boundaries and when | |
70 reading the next mail it uses the previously read data in buffer | |
10479 | 71 - Add some kind of checksum about data+metadata and use it when checking |
72 consistency | |
10798 | 73 - figure out a way to efficiently trigger purging when user has too much |
74 mail expunged (e.g. keep track of total storage size, trigger purging | |
75 when it's 2*quota limit) | |
76 - keep track of total bytes in dbox storage in map header. also if | |
77 possible keep track of refcount=0 bytes. use these to optimize checks. | |
78 - save some stuff to map index header so we don't need to keep retrying | |
79 it. like when saving the lowest file_id which to bother checking. | |
80 - test crash-fixing | |
81 - mail_index_fsck() should perhaps cause dbox to be resynced? | |
10968 | 82 - optimize away reading file header? |
10479 | 83 - maildir: out-of-disk-space failures apparently cause all kinds of |
84 problems, e.g. "Expunged message reappeared", "Duplicate file entry"? | |
10163 | 85 - deliver -r <address> used as autoreplies' From-address? |
86 - add fd limit checks/warnings | |
10546 | 87 - istream-seekable is inefficient. it shouldn't be reading the temp file |
88 immediately after writing to it | |
10968 | 89 - config process is handling requests too slowly. maybe add some caching. |
90 - maybe config should return all of the protocol/local/remote overrides | |
91 when requested? then the caller could do a single lookup at start and | |
92 merge them later internally. this would really help login processes. | |
93 - ipv6: auth penalty should begin from /64 and gradually grow to /48 if | |
94 necessary. and the same could be done for ipv4 as well.. | |
10052 | 95 |
9559 | 96 - dsync: |
11683 | 97 - check for uid conflicts in the middle of mailbox based on next existing |
98 message, not previous | |
10479 | 99 - subscriptions syncing doesn't handle mixed hierarchy separators |
10798 | 100 - Can't rename mailbox INBOX_019acf2169c5784b307f000074ccac23 to INBOX: Target mailbox already exists |
11683 | 101 - copying can't work in non-full mode unless source mailbox is modified |
102 in some way. possibly add some COPY records to transaction log in future? | |
103 - superfast mode: when syncing changes within a mailbox, use modseqs to | |
104 figure out what has changed similar to QRESYNC. | |
10052 | 105 |
10479 | 106 - lda: add some -h 'Header: value' parameter that adds/replaces header |
10163 | 107 - ldap: fix multiple-gid support somehow |
9459 | 108 - search: use mail_get_parts() only when it's already cached. if it's not, |
109 add it to cache afterwards. | |
9235
2e2b957f1cca
Implemented anvil service, which is used to implement mail_max_userip_connections.
Timo Sirainen <tss@iki.fi>
parents:
9159
diff
changeset
|
110 - move ssl proxying code to lib-master |
10798 | 111 - proxy: verify ssl cert's cname |
9004
a5555606846b
Fixed plugins to work with config rewrite.
Timo Sirainen <tss@iki.fi>
parents:
9002
diff
changeset
|
112 - dict pooling |
9015 | 113 |
9000 | 114 /* currently non-external transactions can be applied multiple times, |
115 causing multiple increments. */ | |
116 //FIXME:i_assert((t->flags & MAIL_INDEX_TRANSACTION_FLAG_EXTERNAL) != 0); | |
117 ^ appears to work now though, probably because of the added syncing stuff.. | |
118 | |
119 - transaction log corruption should make sure dovecot.index is rewritten | |
120 and perhaps not delete the file. | |
121 - use backup index in mail_index_fsck() | |
122 | |
8564
3c4934783aff
Mention SSL/TLS in "plaintext auth disallowed" error.
Timo Sirainen <tss@iki.fi>
parents:
8555
diff
changeset
|
123 - proxying: support fallbacking to local (or other?) server if the first |
3c4934783aff
Mention SSL/TLS in "plaintext auth disallowed" error.
Timo Sirainen <tss@iki.fi>
parents:
8555
diff
changeset
|
124 one is down |
8555 | 125 - i_panic("Message count decreased") happens - why? |
8714
bbdbab5354d6
Implemented support for ESORT capability.
Timo Sirainen <tss@iki.fi>
parents:
8564
diff
changeset
|
126 - at least one backtrace shows client_destroy -> client_command_cancel -> |
bbdbab5354d6
Implemented support for ESORT capability.
Timo Sirainen <tss@iki.fi>
parents:
8564
diff
changeset
|
127 imap_sync_deinit |
8555 | 128 - fsck -> log_file_tail_offset 2273345664 -> 996 -> |
129 mail-transaction-log.c: line 341 (mail_transaction_log_set_mailbox_sync_pos): | |
130 assertion failed: (file_offset >= log->head->saved_tail_offset) | |
11765 | 131 - virtual: If last message matching INTHREAD rule gets expunged, the rest of |
132 the thread doesn't go away | |
8488 | 133 - how do shared mailboxes work with plugins? |
134 - lazy-expunge, fts, etc.? | |
11974 | 135 - listescape+acl can't handle shared mailboxes with escape chars |
8488 | 136 - dovecot-acl-list: |
137 - how does it work with global acls? | |
138 - update immediately after SETACL: add/remove entries, update timestamps | |
139 - read the entire file to memory only once and keep it there, stat() later | |
140 to see if it has changed. if not, perhaps don't even bother stat()ing | |
141 dovecot-acl files? at least not that often.. | |
142 - fs quota: getquotaroot inbox vs. other-box should return different quotas | |
143 if two quotas are defined | |
8158 | 144 - new primes code: are hash tables now being resized too often? |
145 - auth_log_prefix setting similar to mail_log_prefix | |
146 | |
147 - thread indexes: if we expunge a duplicate message-id: and we have a sibling | |
148 with identical message-id:, we can probably just move the children? | |
149 (unless there are non-sibling duplicates) | |
150 - SEARCH INTHREAD requires no thread sorting by date - don't do it | |
151 - CONDSTORE: use per-flag/per-keyword conflict checking | |
152 - QRESYNC: Drop expunges from the middle of given seq sets if possible | |
153 - use universal hash functions? | |
154 | |
155 - UIDVALIDITY changed while saving -> sync errors | |
156 - mbox: copy to Trash, manually delete copied msg, change uidvalidity, | |
157 set nextuid=1, copy again -> error | |
158 - recent_uids assert at least with mbox | |
159 - quota fs: Should values returned by quota be divided by the actual | |
160 filesystem block size instead of hardcoded DEV_BSIZE? not with AIX.. | |
7259 | 161 - lucene: handle replacement chars? |
162 - squat: | |
7380 | 163 - wrong indexid |
164 - fts_build_init() assertion failed: (last_uid < last_uid_locked) | |
7259 | 165 - is locking done right? it reads header without file being locked? |
166 - split after ~8 bytes? | |
7391 | 167 - expunges are delayed until more mails are added |
7259 | 168 - test replacement chars (SEARCH / SORT / Squat) |
6987 | 169 |
7449 | 170 - DEBUG: buffer overflow checking code probably doesn't handle a successful |
171 t_try_realloc() or pool_alloconly_realloc() properly | |
8738 | 172 - ldap: |
13324 | 173 - sasl bind + auth_bind=yes should probably be doing only sasl binds.. |
12400 | 174 - support multiple connections for doing auth binds |
8738 | 175 - domain lookups which set the base for user lookup |
176 - same attribute can't be used for multiple values. | |
177 - multiple attributes can't be merged to a single value. | |
12400 | 178 - multiple value could be joined with specified separator (per-field) |
8738 | 179 - implement something like: |
180 user_attrs { | |
181 uid = %{ldap:uidNumber} | |
182 home = %{ldap:homeDirectory} | |
183 quota_bytes = *:bytes=%{ldap:quota} | |
184 } | |
5621 | 185 |
9159
6324a79d3ee1
Initial commit for v2.0 master rewrite. Several features are still missing.
Timo Sirainen <tss@iki.fi>
parents:
9015
diff
changeset
|
186 - maildir+pop3 fast updates: |
6105 | 187 - with locking enabled, pop3 could just keep the one and same sync lock and |
188 do the whole thing using sync transaction | |
189 - don't update dovecot-uidlist if dovecot.index.cache doesn't exist / | |
190 there's nothing to cache | |
191 - if all messages are expunged and there are no unknown extensions in index, | |
192 unlink dovecot.index and rotate log and add some initial useful info to | |
193 the log (uidvalidity, nextuid) | |
5621 | 194 |
6105 | 195 - maildir |
6987 | 196 - don't allow more than 26 keywords |
5621 | 197 |
5557 | 198 - file_cache: we're growing the mmap in page size blocks, which is horribly |
199 slow if mremap() doesn't exist. | |
200 | |
3087 | 201 - keywords: |
202 - add some limits to how many there can be | |
203 - don't return \* in PERMANENTFLAGS when we're full | |
204 - remove unused keywords? | |
2774 | 205 |
11273 | 206 - mail caching |
3904 | 207 - force bits should be used only for nonregistered fields |
208 - change envelope parsing not to use get_headers() so imap.envelope can | |
209 actually be cached without all the headers.. | |
6105 | 210 - if there's no other pressure for compression, we should do it when |
211 enough temp fields are ready to be dropped | |
3904 | 212 - we could try compressing same field values into a single |
213 location in cache file. | |
7259 | 214 - place some maximum limit of fields to cache file? maybe some soft and |
215 hard limits, so when soft limit is reached drop fields that have | |
216 been used only once. when hard limit is reached drop any fields to get | |
217 more space. all this to avoid cache file growing infinitely. | |
2774 | 218 |
3087 | 219 - mbox |
11273 | 220 - UID renumbering doesn't really work after all? |
12476 | 221 - still problems with CRLF mboxes.. especially with broken Content-Length |
222 headers (pointing between CR-LF?) | |
3904 | 223 - syncing existing indexes takes 4x longer than creating new one, why? |
224 - how well does dirty sync + status work? it reads the last mail every | |
225 time? not very good.. | |
226 - always add empty line. make the parser require it too? syncing should | |
227 make sure there always exists two LFs at end of file. raw-mbox-stream | |
228 should make sure the last message ends with LF even if it doesn't exist | |
229 in the file | |
230 - Quote "From ", unquote ">From " | |
4449 | 231 - COPY doesn't work to itself (lock assert crash, for now just disallowed) |
2143 | 232 |
233 - index | |
3087 | 234 - read-only support for mailboxes where we don't have write-access |
6105 | 235 - index file format changes: |
10968 | 236 - split to "old" and "new" indexes and try to avoid loading "old" into |
237 memory until needed | |
6105 | 238 - pack UIDs to beginning of file with UID ranges |
239 - use squat-like compressed uid ranges everywhere | |
6782 | 240 - write first extension intros in dovecot.index.log always with names |
241 - or better yet, drop the intro concept completely as it is now | |
242 - add "transaction boundaries" so we know which records belong to a | |
243 single transaction. | |
244 - only after that we can remove the transaction log offset | |
245 overwriting (otherwise we can get partial transactions in views) | |
2143 | 246 |
3087 | 247 - login |
248 - Digest-MD5: support integrity protection, and maybe crypting. Do it | |
249 through login process like SSL is done? | |
474 | 250 |
3087 | 251 - auth |
4449 | 252 - with blocking passdb we're not caching lookups if the password was wrong |
5557 | 253 - non-plaintext authentication doesn't support all features: |
254 - multiple passdbs don't work, only the first one is used | |
255 - auth cache's last_success password change check doesn't exist | |
9459 | 256 - auth_cache_negative_ttl doesn't check password mismatches |
4449 | 257 - dovecot-auth should limit how fast authentication requests are allowed |
258 from login processes. especially if there's one login/connection the speed | |
3087 | 259 should be something like once/sec. also limit how fast to accept new |
260 connections. | |
261 - support read-only logins. user could with alternative password get only | |
262 read-access to mails so mails could be read relatively safely with | |
263 untrusted computers. Maybe always send [ALERT] about the previous | |
264 read-only login time with IP? | |
265 | |
266 - ssl | |
267 - add setting: ssl_options = bitmask. by default we enable all openssl | |
268 workarounds, this could be used to disable some of them | |
269 - gnutls support isn't working | |
965 | 270 |
3087 | 271 - search |
6232 | 272 - message header search: we should ignore LWSP between two MIME blocks(?) |
273 - message_search_init() could accept multiple search keywords so we | |
1250 | 274 wouldn't need to call it separately for each one (so we wouldn't need |
275 to parse the message multiple times). | |
276 - Create our own extension: When searching with TEXT/BODY, return | |
277 the message text surrounding the keywords just like web search engines | |
278 do. like: SEARCH X-PRINT-MATCHES TEXT "hello" -> * SEARCH 1 "He said: | |
279 Hello world!" 2 "Hello, I'm ...". This would be especially useful with | |
280 the above attachment scanning. | |
0 | 281 |
3087 | 282 - general |
283 - things break if next_uid gets to 2^32 |