From: Rick James
Date: August 31 2011 8:28pm
Subject: Re: Binlog has bad magic number error
List-Archive: http://lists.mysql.com/internals/38377
Message-Id: <4E5E9974.4060502@yahoo-inc.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

What version?  A long time ago, MySQL would briefly swap two file 
descriptors, thereby leading to random corruption.

On 8/31/11 12:05 AM, Mats Kindahl wrote:
> On 08/30/2011 06:21 AM, Kamal O wrote:
>> Hi Mats,
>> Thanks for the reply!
>>
>> I wonder, what could be the resolution for the error.
> There is really no automatic resolution. The magic number can be wrong
> because the file is corrupt, in which case you have to repair it, or
> because it is not a binlog file (did you accidentally copy another
> file on top of the binlog file?)
>
>>
>> Whether it works fine, if slave is restarted pointing to the next
>> binlog file, initial position.
> Is this a question? You get the magic number error when you start at
> the initial position in a binlog file and the magic number is bad. You
> can point replication to any binlog file you like, but if you skip
> events this way, you will (of course) lose changes done at the master.
>
> The decision if these events can be skipped is a decision you have to
> make.
>
> Best wishes,
> Mats Kindahl
>
>>
>>
>> Thanks,
>> Kamal
>>
>>
>>
>> On Mon, Aug 29, 2011 at 12:04 PM, Mats Kindahl
>> <mats.kindahl@stripped<mailto:mats.kindahl@stripped>>  wrote:
>>
>>      Hi Kamalakar,
>>
>>      If you get that error, it is simply a bad magic number for the file
>>      (this is the 4 first bytes of the file), use hexdump to see that the
>>      first bytes are FE 62 69 6E:
>>
>>         $ hexdump -C -n16 context-bin.000001
>>         00000000  fe 62 69 6e 07 a9 53 4e  0f 01 00 00 00 66 00 00
>>         |.bin..SN.....f..|
>>         00000010
>>
>>      There is no connection between the latter error and the first
>>      one. As
>>      a matter of fact, it should not be possible to get the "Can't find
>>      record" if the magic number is not correct for the file and the only
>>      way I can figure out that could give you the "bad magic" error when
>>      skipping events is if the binlog file ends after that and you rotate
>>      to a corrupt file.
>>
>>      Just my few cents,
>>      Mats Kindahl
>>
>>      On 08/19/2011 12:53 PM, Kamalakar Reddy Y wrote:
>>      >  Thanks! Please help! We are getting the below replication error
>>      >
>>      >  Last_SQL_Errno: 1593 Last_SQL_Error: *Error initializing relay
>>      log position:
>>      >  Binlog has bad magic number;  It's not a binary log file that
>>      can be used by
>>      >  this version of MySQL.*
>>      >
>>      >
>>      >
>>      >  Initially we were facing the below error, which we skipped
>>      that resulted the
>>      >  above error.
>>      >
>>      >
>>      >
>>      >  Last_Error: Could not execute Update_rows event on tables
>>      xxxxxxx.zzzzz:
>>      >  Can't find record in 'zzzzz', Error_code:1032; handle error
>>      >  HA_ERR_KEY_NOT_FOUND; the event's master log
>>      mysql-binlog-LLL01.000683,
>>      >  end_log_pos 776331958.
>>      >
>>      >
>>      >
>>      >  Please do help!......
>>      >
>>
>>
>

-- 
Rick James - MySQL Geek