Index: [Article Count Order] [Thread]

Date:  Tue, 17 Mar 2009 08:27:20 -0600
From:  "Gerald Waugh" <gwaugh (at mark) frontstreetnetworks.com>
Subject:  [coba-e:15347] Re: Re  Broken FTP
To:  <coba-e (at mark) bluequartz.org>
Message-Id:  <85B65413FDF8424CA9ABEEB812CF36D7@raqware>
References:  <45DACA7BE1DEA245AF6C95E5B94E531B0D07A62D72 (at mark) datastore.mainline.local> <49BF9787.5030804 (at mark) virtbiz.com> <45DACA7BE1DEA245AF6C95E5B94E531B0D07A62D82 (at mark) datastore.mainline.local>
X-Mail-Count: 15347

Colin Jack wrote; Tuesday, March 17, 2009 7:55 AM

>> > * The FTP server is operating normally.
>> >
>> > Seeing this about four or five times in 24 hours.
>>
>> We have yet to witness this on any server and we run service-level
>> monitoring on every one of our hosting boxes at 5 minute intervals.
>>
>> This sort of smells DNS'y to me with the off/on-again nature that you're
>> describing.  You're not running DNS lookups on your FTP, are you?   How
>> is your DNS recursion doing?   Does /var/log/messages show anything odd
>> w/ regard to proftpd?
>>
>> --
>> Chris Gebhardt wrote;
>
> These servers have been rock solid for years (apart from the BO one) ... 
> not running anything like DNS lookups.
> Absolutely nothing has changed other than the yum update which updated 
> ProFTPD.
> The servers are colocoed at two different data centres.
>
> Nope nothing untoward in the logs.

Look at /usr/sausalito/swatch/bin/am_ftp.exp
And try to run the telnet commands manually and see if you get errors.
   telnet localhost 21
Then expect response;
  "Connected to"
Then expect response;
  "220 "
Then send;
  QUIT or quit
Then expect response;
  "221 "
This should all work.
Then execute;
   swatch | cat /var/log/messages | grep cced | grep ftp

We have found that several servers we configured and updated give 'Active 
Monitor' errors, but FTP is actually working.
It may be a timeout issue. Strangely, the Stongbolt servers did not have 
this issue.

Gerald