495884965

今天log出現下面message
Sep 25 23:52:16  dovecot: POP3(user02): rename(/user02/new/1285097266.28195_1.prion, /vuser02/cur/1285097266.28195_1.prion:2,) failed: Stale NFS file handle

辜狗到下面的文章

http://www.myfaq.com.cn/2005September/2005-09-13/202824.html

“ 當我們已經mount 上的file or directory ,在server上突然被remove or unexport ,就會出現此訊息“。
就是說，如果client端mount上了server端的directory之後，如果server端又將這個directory unshare了，那麼就會在client端出現這個錯誤。

其實解決辦法很簡單，就是在client上把那個檔unmount掉。
可是其實做起來很難，因為經常會得到 is busy的錯誤。

可以用下面的三個方法：
1)用fuser殺掉佔用那個目錄的Process
fuser -k /directory

2)在啟動的Process裏面查找
ps -ef |awk '{print }' |grep -v PID |xargs /usr/proc/bin/pwdx 


3)umount -f /directory
強制umount

不過umount -f 不太有用，查man似乎有什麼限制(沒仔細看)

所以還是用fuser搞定它!!IMBA

======================================
dovecot: POP3(user1): Couldn't open INBOX: Timeout while waiting for lock

參考

主要是說因為/etc/dovecot.conf裡面

pop3_lock_session = yes

What it means is that you have this enabled and user tries to log in
with POP3 while another POP3 session already exists and isn't dying
after 5 minutes. Some possible reasons:
 - User using multiple POP3 clients?
 - User using dialup and connection dies while downloading messages?
 - User has huge mailboxes that take a long time to download?
In any case setting pop3_lock_session=no gets rid of these errors,
although that then violates POP3 RFC.