MySQL Bugs: #35126: myisamchk: Disk is full writing zzz.MYI

Bug #35126	myisamchk: Disk is full writing zzz.MYI
Submitted:	6 Mar 2008 22:57	Modified:	15 Nov 2008 18:46
Reporter:	Oleg Alexeenkov	Email Updates:
Status:	No Feedback	Impact on me:	None
Category:	MySQL Server: MyISAM storage engine	Severity:	S3 (Non-critical)
Version:	5.1.23-rc FreeBSD port: mysql-server-5.1	OS:	FreeBSD (7.0)
Assigned to:	Assigned Account	CPU Architecture:	Any
Tags:	disk, full, MyISAMchk, Writing

Description:
i run 
myisamchk -v --recover --force /var/db/mysql/z/*.MYI
but during this process (~30G tables, ~40G expected index) ~4-8 hours
free space was ended and i recieve lot of messages:

2008/03/06-23:40:18 [11041] dbg - Fixing index 10
2008/03/06-23:40:18 [11041] dbg   - Searching for keys, allocating buffer for 61946206 keys
2008/03/06-23:40:18 [11041] dbg   - Last merge and dumping keys
2008/03/06-23:40:18 [11041] dbg myisamchk: Disk is full writing '/storage/2/db/z/z.MYI' (Errcode: 28). Waiting for someone to free space... Retry in 60 secs
/storage/2: write failed, filesystem is full

then i clean disk:
Filesystem     Size    Used   Avail Capacity  Mounted on
/dev/ad6s1d    451G    249G    166G    60%    /storage/2

but myisamchk is staying 

  PID USERNAME       THR PRI NICE   SIZE    RES STATE  C   TIME   WCPU COMMAND
11312 root             1  -4    0  3088M  3087M snaplk 2 388:14  0.00% myisamchk

and continue write me

2008/03/07-01:25:15 [11041] dbg myisamchk: Disk is full writing '/storage/2/db/proiska/filebase.MYI' (Errcode: 28). Waiting for someone to free space... Retry in 60 secs
2008/03/07-01:35:54 [11041] dbg myisamchk: Disk is full writing '/storage/2/db/proiska/filebase.MYI' (Errcode: 28). Waiting for someone to free space... Retry in 60 secs

How to repeat:
big table, low disk space
myisamchk -v --recover --force /var/db/mysql/z/*.MYI

second retry:

2008/03/07-11:24:52 [30324] dbg - Fixing index 10
2008/03/07-11:24:52 [30324] dbg   - Searching for keys, allocating buffer for 61946169 keys
2008/03/07-11:24:52 [30324] dbg myisamchk: Disk is full writing '/storage/2/tmp/STc7nsda' (Errcode: 28). Waiting for someone to free space... Retry in 60 secs

/storage/2: write failed, filesystem is full

but after cleaning myisamchk starts working.
  PID USERNAME       THR PRI NICE   SIZE    RES STATE  C   TIME   WCPU COMMAND
30590 root             1 120    0  3088M  3087M CPU1   0 510:16 98.68% myisamchk

Many thanks for writing a bug report.
How does your indexes look like?

CREATE TABLE `filebase` (
 `prot` varchar(6) NOT NULL DEFAULT '',
 `host` varchar(64) NOT NULL DEFAULT '',
 `path` varchar(182) NOT NULL DEFAULT '',
 `name` varchar(182) NOT NULL DEFAULT '',
 `ext` varchar(64) NOT NULL DEFAULT '',
 `size` bigint(20) unsigned NOT NULL DEFAULT '0',
 `time` int(10) unsigned NOT NULL DEFAULT '0',
 `desc` varchar(200) NOT NULL DEFAULT '',
 `scan` int(10) unsigned NOT NULL DEFAULT '0',
 `added` int(10) unsigned NOT NULL DEFAULT '0',
 `stem` varchar(128) NOT NULL DEFAULT '',
 PRIMARY KEY (`prot`,`host`,`path`,`name`,`ext`),
 KEY `prot_i` (`prot`(3)),
 KEY `host_i` (`host`(16)),
 KEY `path_i` (`path`(128)),
 KEY `name_i` (`name`(32)),
 KEY `ext_i` (`ext`(8)),
 KEY `size_i` (`size`),
 KEY `time_i` (`time`),
 KEY `scan_i` (`scan`),
 FULLTEXT KEY `stemi` (`stem`),
 FULLTEXT KEY `fulltexti` (`name`,`ext`,`desc`)
) ENGINE=MyISAM DEFAULT CHARSET=cp1251

Thank you for the feedback.

But you said myisamchk stops working if no space left on device and starts after cleaning. So what the bug is?

it stops totally if space ends during this stage:
2008/03/06-23:40:18 [11041] dbg   - Last merge and dumping keys

but continue working after cleaning if space ends on this stage:
2008/03/06-23:40:18 [11041] dbg   - Searching for keys, allocating buffer for 61946206 keys

Thank you for the feedback.

I can not repeat described behavior with latest version. Please try with it. Also please check with myisamchk from our distribution available from dev.mysql.com/downloads to be sure this is not a problem of FreeBSD port.

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".