Re: Problem with included global sieve script when mailbox contains a dot
Can anyone comment on that ? We might be doing something wrong while settings our ACLs, but looking with wireshark at how cyradm I think we're fine and that there's really a bug in here. Thomas, Blue Mind. On Mon, Nov 19, 2012 at 8:39 PM, Thomas Cataldo tcata...@gmail.com wrote: Hi, I did some more research on my problem and tracked the problem down to this test in append.c (cyrus 2.4) : else if (!strcmp(flag[i], \\flagged)) { if (as-myrights ACL_WRITE) { // this test fails when mailbox is a.b@domain message_index.system_flags |= FLAG_FLAGGED; } } My problem is probably related to unixhierarchysep settings and the way ACLs are set, but the following patch fixes it : diff --git a/imap/append.c b/imap/append.c index d3c2bd4..b622dd1 100644 --- a/imap/append.c +++ b/imap/append.c @@ -160,12 +160,25 @@ int append_setup(struct appendstate *as, const char *name, { int r; struct quota q; +unsigned int pos; +size_t at_char; +char* fixed_acls; as-mailbox = NULL; r = mailbox_open_iwl(name, as-mailbox); if (r) return r; -as-myrights = cyrus_acl_myrights(auth_state, as-mailbox-acl); +fixed_acls = strdup(as-mailbox-acl); +syslog(LOG_ERR, before acls tweaks: %s\n, fixed_acls); +at_char = strcspn(fixed_acls, @); +for (pos=0; pos at_char; pos++) { + if (fixed_acls[pos] == '.') { + fixed_acls[pos] = '^'; + } +} +syslog(LOG_ERR, after acls tweaks: %s\n, fixed_acls); + +as-myrights = cyrus_acl_myrights(auth_state, fixed_acls); if ((as-myrights aclcheck) != aclcheck) { r = (as-myrights ACL_LOOKUP) ? The two logs I added give the following output : Nov 19 20:37:00 ppday cyrus/lmtp[42207]: before acls tweaks: a...@buffy.vmw #011lrswipkxtecda#011admin0#011lrswipkxtecda#011ad...@buffy.vmw #011lrsp#011 Nov 19 20:37:00 ppday cyrus/lmtp[42207]: after acls tweaks: a^b...@buffy.vmw #011lrswipkxtecda#011admin0#011lrswipkxtecda#011ad...@buffy.vmw #011lrsp#011 I hope this test patch might help figure out what is wrong. My acls setup or a bug. Regards, Thomas Cataldo, Blue Mind. On Wed, Nov 14, 2012 at 10:40 AM, Thomas Cataldo tcata...@gmail.comwrote: Hello, Consider the following sieve setup : root@ppday:/var/spool/sieve# cat global/buffy.vmw.sieve.script require [body, copy, fileinto, imapflags ]; # filters if allof ( address :contains from admin) { setflag \\Flagged; } # END root@ppday:/var/spool/sieve# cat domain/b/buffy.vmw/a/a^b/bm-user5.sieve.script require [body, copy, fileinto, imapflags, vacation, include ]; include :global buffy.vmw.sieve; # END root@ppday:/var/spool/sieve# cat domain/b/buffy.vmw/s/sp/bm-user6.sieve.script require [body, copy, fileinto, imapflags, vacation, include ]; include :global buffy.vmw.sieve; # END When ad...@buffy.vmw sends an email to a...@buffy.vmw and s...@buffy.vmw, the following happens : - the mail is correctly flagged for sp - no flagging for a.b Nothing interesting in my logs when the mail is delivered : Nov 14 10:17:20 ppday cyrus/master[19560]: about to exec /usr/lib/cyrus/bin/lmtpd Nov 14 10:17:20 ppday cyrus/lmtp[19560]: executed Nov 14 10:17:20 ppday cyrus/lmtp[19560]: skiplist: checkpointed /var/lib/cyrus/statuscache.db (0 records, 144 bytes) in 0 seconds Nov 14 10:17:21 ppday cyrus/lmtp[19560]: accepted connection Nov 14 10:17:21 ppday cyrus/lmtp[19560]: connection from ppday.buffy.vmw [172.16.78.128] preauth'd as postman Nov 14 10:17:21 ppday cyrus/lmtp[19560]: Delivered: 1c42d02ef8953cb95219dafba...@buffy.vmw to mailbox: buffy.vmw!user.a^b Nov 14 10:17:21 ppday cyrus/lmtp[19560]: USAGE a^b user: 0.004000 sys: 0.008000 Nov 14 10:17:21 ppday cyrus/lmtp[19560]: Delivered: 1c42d02ef8953cb95219dafba...@buffy.vmw to mailbox: buffy.vmw!user.admin Nov 14 10:17:21 ppday cyrus/lmtp[19560]: USAGE admin user: 0.00 sys: 0.004000 Nov 14 10:17:21 ppday cyrus/lmtp[19560]: Delivered: 1c42d02ef8953cb95219dafba...@buffy.vmw to mailbox: buffy.vmw!user.sp Nov 14 10:17:21 ppday cyrus/lmtp[19560]: USAGE sp user: 0.00 sys: 0.00 Nov 14 10:17:21 ppday postfix/lmtp[19551]: 5DA0A21E37: to=a...@buffy.vmw, relay=172.16.78.128[172.16.78.128]:2400, delay=0.93, delays=0.07/0.03/0.03/0.79, dsn=2.1.5, status=sent (250 2.1.5 Delivery OK) Nov 14 10:17:21 ppday postfix/lmtp[19551]: 5DA0A21E37: to=ad...@buffy.vmw, relay=172.16.78.128[172.16.78.128]:2400, delay=0.93, delays=0.07/0.03/0.03/0.79, dsn=2.1.5, status=sent (250 2.1.5 Delivery OK) Nov 14 10:17:21 ppday postfix/lmtp[19551]: 5DA0A21E37: to=s...@buffy.vmw, relay=172.16.78.128[172.16.78.128]:2400, delay=0.93, delays=0.07/0.03/0.03/0.79, dsn=2.1.5, status=sent (250 2.1.5 Delivery OK) Nov 14 10:17:21 ppday postfix/qmgr[20160]: 5DA0A21E37: removed Any know bugs with global sieve vs mailboxes with dot in them ? Can I enable more verbose
Re: Problem with included global sieve script when mailbox contains a dot
Sorry I haven't commented. Yes, I think you're right - it's a bug. I tried to fix something like it a while back and created more and worse bugs, so I reverted the change. The way that usernames and folder details are handled in sieve is a right mess... conversions backwards and forwards between internal and external namespaces. It's on the to look at before 2.5 list. I'm travelling at the moment, so I'm not spending as much time on Cyrus as I'd like :( I'd love anyone else who has time to take a look too... see if they can untangle the conversions! Bron. On Sun, Nov 25, 2012, at 11:31 AM, Thomas Cataldo wrote: Can anyone comment on that ? We might be doing something wrong while settings our ACLs, but looking with wireshark at how cyradm I think we're fine and that there's really a bug in here. Thomas, Blue Mind. On Mon, Nov 19, 2012 at 8:39 PM, Thomas Cataldo [1]tcata...@gmail.com wrote: Hi, I did some more research on my problem and tracked the problem down to this test in append.c (cyrus 2.4) : else if (!strcmp(flag[i], \\flagged)) { if (as-myrights ACL_WRITE) { // this test fails when mailbox is a.b@domain message_index.system_flags |= FLAG_FLAGGED; } } My problem is probably related to unixhierarchysep settings and the way ACLs are set, but the following patch fixes it : diff --git a/imap/append.c b/imap/append.c index d3c2bd4..b622dd1 100644 --- a/imap/append.c +++ b/imap/append.c @@ -160,12 +160,25 @@ int append_setup(struct appendstate *as, const char *name, { int r; struct quota q; +unsigned int pos; +size_t at_char; +char* fixed_acls; as-mailbox = NULL; r = mailbox_open_iwl(name, as-mailbox); if (r) return r; -as-myrights = cyrus_acl_myrights(auth_state, as-mailbox-acl); +fixed_acls = strdup(as-mailbox-acl); +syslog(LOG_ERR, before acls tweaks: %s\n, fixed_acls); +at_char = strcspn(fixed_acls, @); +for (pos=0; pos at_char; pos++) { + if (fixed_acls[pos] == '.') { + fixed_acls[pos] = '^'; + } +} +syslog(LOG_ERR, after acls tweaks: %s\n, fixed_acls); + +as-myrights = cyrus_acl_myrights(auth_state, fixed_acls); if ((as-myrights aclcheck) != aclcheck) { r = (as-myrights ACL_LOOKUP) ? The two logs I added give the following output : Nov 19 20:37:00 ppday cyrus/lmtp[42207]: before acls tweaks: a...@buffy.vmw#011lrswipkxtecda#011admin0#011lrswipkxtecda#011admin@buff y.vmw#011lrsp#011 Nov 19 20:37:00 ppday cyrus/lmtp[42207]: after acls tweaks: a^b...@buffy.vmw#011lrswipkxtecda#011admin0#011lrswipkxtecda#011admin@buff y.vmw#011lrsp#011 I hope this test patch might help figure out what is wrong. My acls setup or a bug. Regards, Thomas Cataldo, Blue Mind. On Wed, Nov 14, 2012 at 10:40 AM, Thomas Cataldo [2]tcata...@gmail.com wrote: Hello, Consider the following sieve setup : root@ppday:/var/spool/sieve# cat global/buffy.vmw.sieve.script require [body, copy, fileinto, imapflags ]; # filters if allof ( address :contains from admin) { setflag \\Flagged; } # END root@ppday:/var/spool/sieve# cat domain/b/buffy.vmw/a/a^b/bm-user5.sieve.script require [body, copy, fileinto, imapflags, vacation, include ]; include :global buffy.vmw.sieve; # END root@ppday:/var/spool/sieve# cat domain/b/buffy.vmw/s/sp/bm-user6.sieve.script require [body, copy, fileinto, imapflags, vacation, include ]; include :global buffy.vmw.sieve; # END When ad...@buffy.vmw sends an email to a...@buffy.vmw and s...@buffy.vmw, the following happens : - the mail is correctly flagged for sp - no flagging for a.b Nothing interesting in my logs when the mail is delivered : Nov 14 10:17:20 ppday cyrus/master[19560]: about to exec /usr/lib/cyrus/bin/lmtpd Nov 14 10:17:20 ppday cyrus/lmtp[19560]: executed Nov 14 10:17:20 ppday cyrus/lmtp[19560]: skiplist: checkpointed /var/lib/cyrus/statuscache.db (0 records, 144 bytes) in 0 seconds Nov 14 10:17:21 ppday cyrus/lmtp[19560]: accepted connection Nov 14 10:17:21 ppday cyrus/lmtp[19560]: connection from ppday.buffy.vmw [172.16.78.128] preauth'd as postman Nov 14 10:17:21 ppday cyrus/lmtp[19560]: Delivered: 1c42d02ef8953cb95219dafba...@buffy.vmw to mailbox: buffy.vmw!user.a^b Nov 14 10:17:21 ppday cyrus/lmtp[19560]: USAGE a^b user: 0.004000 sys: 0.008000 Nov 14 10:17:21 ppday cyrus/lmtp[19560]: Delivered: 1c42d02ef8953cb95219dafba...@buffy.vmw to mailbox: buffy.vmw!user.admin Nov 14 10:17:21 ppday cyrus/lmtp[19560]: USAGE admin user: 0.00 sys: 0.004000 Nov 14 10:17:21 ppday cyrus/lmtp[19560]: Delivered: 1c42d02ef8953cb95219dafba...@buffy.vmw to mailbox: buffy.vmw!user.sp Nov 14 10:17:21 ppday cyrus/lmtp[19560]: USAGE sp user: 0.00 sys: 0.00 Nov 14 10:17:21 ppday postfix/lmtp[19551]: 5DA0A21E37: to=a...@buffy.vmw, relay=172.16.78.128[172.16.78.128]:2400, delay=0.93, delays=0.07/0.03/0.03/0.79, dsn=2.1.5, status=sent (250 2.1.5 Delivery