[Dovecot] NFS error, device busy

[Grandy Fu]

>
> On 2/16/2009 9:50 PM, Grandy Fu wrote:
>   
>> > I am running dovecot 1.1.10 and 1.1.11 on Solaris 10, mailboxes are in
>> > mbox format and served by Solaris 10 NFSv3, the dovecot cache is on
>> > local disk.
>> >
>> > Some of my users always had above nfs error and their mailbox usually
>> > very big, over 100M, 400M, or even 800M.
>>     
>
> Please don't put the entire error message in the subject.
>
>   
Sorry about that.

> Please post output of dovecot -n and some logs exhibiting the problem.

Here is some log entries.

---------------
dovecot: Feb 19 10:07:02 Error: IMAP(user): 
nfs_flush_file_handle_cache_dir: rmdir(/var/mail) failed: Device busy
dovecot: Feb 19 10:07:02 Error: IMAP(user): 
nfs_flush_file_handle_cache_dir: rmdir(/var/mail) failed: Device busy
dovecot: Feb 19 10:07:02 Error: IMAP(user): 
nfs_flush_file_handle_cache_dir: rmdir(/var/mail) failed: Device busy
dovecot: Feb 19 10:07:02 Error: IMAP(user): 
nfs_flush_file_handle_cache_dir: rmdir(/var/mail) failed: Device busy
dovecot: Feb 19 10:07:02 Error: IMAP(user): 
nfs_flush_file_handle_cache_dir: rmdir(/var/mail) failed: Device busy
dovecot: Feb 19 10:07:02 Error: IMAP(user): 
nfs_flush_file_handle_cache_dir: rmdir(/var/mail) failed: Device busy
-------------------------------------
part of dovecot -n output
---------------------------
# 1.1.11: /usr/local/.dovecot1111/etc/dovecot.conf
# OS: SunOS 5.10 sun4u
log_path: /var/log/dovecot
protocols: imaps pop3s imap
listen(default): *:143
listen(imap): *:143
listen(pop3): *
ssl_listen(default): *:993
ssl_listen(imap): *:993
ssl_listen(pop3):
login_dir: /usr/local/.dovecot1111/var/run/dovecot/login
login_executable(default): 
/usr/local/.dovecot1111/libexec/dovecot/imap-login
login_executable(imap): /usr/local/.dovecot1111/libexec/dovecot/imap-login
login_executable(pop3): /usr/local/.dovecot1111/libexec/dovecot/pop3-login
mail_location: 
mbox:~/mail:INBOX=/var/mail/%u:INDEX=/localdisk/dovecot.cache/cache/%n
mail_debug: yes
mmap_disable: yes
dotlock_use_excl: no
mail_nfs_storage: yes
lock_method: dotlock
mbox_read_locks: dotlock
mbox_write_locks: dotlock
mbox_dotlock_change_timeout: 30
plugin:
  quota: fs:INBOX:user:mount=/var/mail
  quota2: fs:home:noenforcing:mount=/home

----------------------------------------------------
The problem happens to users' with their inbox over 100M and it happens 
randomly.

Grandy