Using maildir++ with LDA and POP/IMAP we have been seeing intermittant maildirsize corruption. Where an operation will be prepended by a handful of ASCI null characters ^@^@. Eg: # cat maildirsize 5120000S -46895 -1 93359 1 60576 1 2858 1 49547 1 -206340 -4 94603 1 84785 1 54463 1 82486 1 97104 1 ^@^@^@^@^@^@^@^@40488 1 73397 1 83662 1 24668 1 42673 1 12081 1 -713606 -12 The instances of this that we have seen have all been deliver operations. Anyone else seen this ? Regards __________________________________________ Dean Manners
On Fri, 2007-03-02 at 12:32 +1100, Dean Manners wrote:> Using maildir++ with LDA and POP/IMAP we have been seeing intermittant > maildirsize corruption. Where an operation will be prepended by a handful > of ASCI null characters ^@^@.Is this a real problem? From the code: /* We rely on O_APPEND working in here. That isn't NFS-safe, but it isn't necessarily that bad because the file is recreated once in a while, and sooner if corruption causes calculations to go over quota. This is also how Maildir++ spec specifies it should be done.. */ I guess I could make it recalculate the quota if it detects any errors in the file. -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 189 bytes Desc: This is a digitally signed message part URL: <http://dovecot.org/pipermail/dovecot/attachments/20070306/c6ff3cbe/attachment.bin>
Hi Timo, it would be great if someone can overlook it it seems there are some bugs. If anyone can do a patch, I would test it and make a report. Is there anything like bugzilla where I can make a bugreport ? Thanks, Frank -----Urspr?ngliche Nachricht----- Von: dovecot-bounces at dovecot.org [mailto:dovecot-bounces at dovecot.org] Im Auftrag von Timo Sirainen Gesendet: Dienstag, 6. M?rz 2007 17:21 An: Dean Manners Cc: 'Dovecot Mailing List' Betreff: Re: [Dovecot] maildirsize corruption On Fri, 2007-03-02 at 12:32 +1100, Dean Manners wrote:> Using maildir++ with LDA and POP/IMAP we have been seeing intermittant > maildirsize corruption. Where an operation will be prepended by a > handful of ASCI null characters ^@^@.Is this a real problem? From the code: /* We rely on O_APPEND working in here. That isn't NFS-safe, but it isn't necessarily that bad because the file is recreated once in a while, and sooner if corruption causes calculations to go over quota. This is also how Maildir++ spec specifies it should be done.. */ I guess I could make it recalculate the quota if it detects any errors in the file.