MySQL Bugs: #28154: "Got error 127 when reading table" appears randomly on a big MYISAM table

Bug #28154	"Got error 127 when reading table" appears randomly on a big MYISAM table
Submitted:	29 Apr 2007 17:26	Modified:	9 Nov 2007 18:44
Reporter:	Dmitry Panov	Email Updates:
Status:	No Feedback	Impact on me:	None
Category:	MySQL Server: MyISAM storage engine	Severity:	S3 (Non-critical)
Version:	5.0.32, 5.0.37	OS:	Linux
Assigned to:		CPU Architecture:	Any

Description:
The error message looks as follows:

070422  7:01:07 [ERROR] Got error 127 when reading table
'./intotalnews/Story_Fulltext'

It appears in the log file from time to time on 2 servers running different OS versions, different MySQL versions and having different hardware platform. 

First server is running debian sarge:

Linux hostinghi 2.6.8-2-686-smp #1 SMP Thu May 19 17:27:55 JST 2005 i686
GNU/Linux

glibc 2.3.2

NPTL 0.60

The second server is running debian etch:

Linux hosting3 2.6.18-4-amd64 #1 SMP Wed Feb 21 14:29:38 UTC 2007 x86_64
GNU/Linux

glibc 2.3.6

NPTL 2.3.6

The table in question is quite large (3GB+ data and 4GB+ full-text index). MySQL
5.0.16 (static, downloaded from mysql.com) is running fine on the first server.
However version 5.0.37 (static and glibc 2.3 downloaded from the site) running
on the same server as well as debian etch version (based on 5.0.32) running on
the second server give random "error 127". No crash occurs and a subsequent
query may succeed. 

It is the only MYISAM table in the database (others use InnoDB). 
It also looks like these errors do not always cause the query to fail. But eventually the query fails with "Table 'b' is marked as crashed and should be repaired". Repairing helps for some time, then the error appears again. 

How to repeat:
Don't know

Thank you for a problem report. Please, send my.cnf files content for all instances involved. I am specifically interested in thread_cache_size values.

Added the files. thread_cache_size is 8 in both cases. 

There are 2 important things to note. 

These 2 servers are slaves, the master is running exactly the same hardware and software as server 2. I haven't seen any 127 errors on it. The difference is that no SELECTs occur on the master, just INSERTs.

Another thing is I tried dropping the full-text index from the table and it seemed to help. There were no errors for a couple of days while the index was absent. After creating it again I got the error in just a couple of minutes.

Please, try to repeat with a newer version, 5.0.41, and inform about the results. Check if thread_cache_size=0 will make any difference.

No feedback was provided for this bug for over a month, so it is
being suspended automatically. If you are able to provide the
information that was originally requested, please do so and change
the status of the bug back to "Open".

Sorry for delay... Just checked it with 5.0.45 (glibc2.3 linked), it's the same effect. Also setting thread_cache_size=0 didn't help on either of the servers.

Thank you for the feedback.

Please provide output of SHOW CREATE TABLE and SHOW TABLE STATUS for problem table.

might be related to bug #29838

mysql> SHOW CREATE TABLE Story_Fulltext;
+----------------+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| Table          | Create Table                                                                                                                                                                                                                                                                |
+----------------+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| Story_Fulltext | CREATE TABLE `Story_Fulltext` (
  `Story_Translation_ID` int(11) default NULL,
  `HeadLine` text,
  `Body` text,
  UNIQUE KEY `Story_Fulltext_IDX2` (`Story_Translation_ID`),
  FULLTEXT KEY `Story_Fulltext_IDX1` (`HeadLine`,`Body`)
) ENGINE=MyISAM DEFAULT CHARSET=utf8 |
+----------------+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+

mysql> show table status like 'Story_Fulltext';
+----------------+--------+---------+------------+---------+----------------+-------------+-----------------+--------------+-----------+----------------+---------------------+---------------------+---------------------+-----------------+----------+----------------+---------+
| Name           | Engine | Version | Row_format | Rows    | Avg_row_length | Data_length | Max_data_length | Index_length | Data_free | Auto_increment | Create_time         | Update_time         | Check_time          | Collation       | Checksum | Create_options | Comment |
+----------------+--------+---------+------------+---------+----------------+-------------+-----------------+--------------+-----------+----------------+---------------------+---------------------+---------------------+-----------------+----------+----------------+---------+
| Story_Fulltext | MyISAM |      10 | Dynamic    | 2840730 |           2102 |  5971579192 | 281474976710655 |   3579031552 |         0 |           NULL | 2007-04-29 17:28:36 | 2007-08-14 12:24:49 | 2007-04-29 18:31:06 | utf8_general_ci |     NULL |                |         |
+----------------+--------+---------+------------+---------+----------------+-------------+-----------------+--------------+-----------+----------------+---------------------+---------------------+---------------------+-----------------+----------+----------------+---------+
1 row in set (0.00 sec)

Dmitry, please try with latest version of mysql. I suspect fix for Bug #29838 will solve this report. Thanks,

No feedback was provided for this bug for over a month, so it is
being suspended automatically. If you are able to provide the
information that was originally requested, please do so and change
the status of the bug back to "Open".

I have the same problem on 2 fedora 7 and 8 which are setup as multimaster replication.(mysql 5.0.45)
An error appears every few days/hours, and there seem to be no error in the tables when doing a check table.
When comparing with other cases, it looks like it is related to replication.

Could it be that when the data is being updated from the master, and it is being read, this error can trigger?

joe,

thank you for the feedback. But version 5.0.45 does not contain fix for the bug #29838. Please upgrade to version 5.0.51 or newer (current is 5.0.67) and try with it