Hi,
same here on my installation. Version: Enterprise Edition: 2:2.2.24.1-2
Some logs:
...
Jun 5 07:40:01 dovecot-server dovecot: imap(user@domain.com pid:27937 session:
Got also this errors:
Jun 5 07:40:01 dovecot-server dovecot: imap(user@domain.com pid:27937 session:
...
Thanks Urban
Am 09.06.2016 um 16:06 schrieb Dave:
On 02/06/2016 22:58, Timo Sirainen wrote:
On 01 Jun 2016, at 16:48, Alessio Cecchi alessio@skye.it wrote:
Hi,
after the last upgrade to Dovecot 2.2.24.2 (d066a24) I see an increased number of errors "Broken MIME parts" for users in dovecot log file, here an example:
Jun 01 15:25:29 Error: imap(alessio.cecchi@skye.it): Corrupted index cache file /home/domains/skye.it/alessio.cecchi/Maildir/dovecot.index.cache: Broken MIME parts for mail UID 34 in mailbox INBOX: Cached MIME parts don't match message during parsing: Cached header size mismatch (parts=410000005b070000000000007b07000000000000fc0a000000000000400b0000000000000300000048000000800700000000000060000000000000006400000000000000200000000000000021000000000000000100000040000000260800000000000027000000000000002900000000000000ea00000000000000f000000000000000440000005d090000000000001e000000000000002000000000000000b308000000000000e0080000000000002d00000001000000410000007b09000000000000b208000000000000de080000000000000000000000000000000000000000000000000000)
..
but the error reappears always (for the same UID) when I do "search" from webmail. All works fine for the users but I don't think is good to have these errors in log file.
If it's reproducible for a specific email, can you send me the email?
I'm replying to this again for a couple of reasons:
I've not heard any further discussion and I accidentally replied to the wrong thread initially (oops!)
It's actually looking to become a fairly serious issue (extra 100Mb/s of network traffic, extra 10K NFS ops/sec)
I've been seeing the same problem after upgrading from 2.2.18 to 2.2.24 with identical config. Reads from mailboxes have doubled, which looks consistent with repeated index rebuilds. It's not got any better over time, so the indexes aren't managing to self-heal (over 1300 of these errors today so far, for example).
[2016-06-09T09:50:30+0100] imap(xxxx): Error: Corrupted index cache file /mnt/index/c69/923413/.INBOX/dovecot.index.cache: Broken MIME parts for mail UID 74359 in mailbox INBOX: Cached MIME parts don't match message during parsing: Cached header size mismatch (parts=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 0 000000000000000000000000000000000000000000)
[2016-06-09T10:05:32+0100] imap(xxxx): Error: Corrupted index cache file /mnt/index/498/1603514/.INBOX/dovecot.index.cache: Broken MIME parts for mail UID 810 in mailbox INBOX: Cached MIME parts don't match message during parsing: Cached header size mismatch (parts=41000000dc02000000000000ef02000000000000e8060000000000001c070000000000000300000048000000fc020000000000005b000000000000005f0000000000000098000000000000009b0000000000000003000000400000001004000000000000270000000000000029000000000000000a0100000000000011010000000000004400000062050000000000001e00000000000000200000000000000020040000000000003a040000000000001a000000010000004100000080050000000000001f0400000000000038040000000000000000000000000000000000000000000000000000)
Again, usual force-resync of the index or deletion and recreation don't help, the same indexes keep corrupting based on certain messages.
Checking one of the triggering emails, maildir tags for size and virtual size are correct, contains a 6 part multipart/mixed with the first part being itself a 2 part multipart/alternative, both have different boundary ids (unfortunately I still cannot release the email itself, I'm hoping the error will trigger on some that I can release, none so far)