Bug #68345 drop --innodb_fast_shutdown=2 for first MTR (bootstrap) startup
Submitted: 12 Feb 2013 3:55 Modified: 25 Nov 2013 16:21
Reporter: Roel Van de Paar (OCA) Email Updates:
Status: Closed Impact on me:
None 
Category:Tests: Engine Severity:S7 (Test Cases)
Version:5.x inc 5.6 OS:Any
Assigned to: CPU Architecture:Any
Tags: mtr
Triage: Needs Triage: D4 (Minor)

[12 Feb 2013 3:55] Roel Van de Paar
Description:
MTR, when it executes bootstrap will not drop options like "--innodb_fast_shutdown=2". This works fine for the bootstrap (from bootstrap.log):

===========
2013-02-12 10:08:46 32153 [Note] InnoDB: Setting log file ./ib_logfile101 size to 48 MB
2013-02-12 10:08:46 32153 [Note] InnoDB: Setting log file ./ib_logfile1 size to 48 MB
2013-02-12 10:08:47 32153 [Note] InnoDB: Renaming log file ./ib_logfile101 to ./ib_logfile0
2013-02-12 10:08:47 32153 [Warning] InnoDB: New log files created, LSN=45781
2013-02-12 10:08:47 32153 [Note] InnoDB: Doublewrite buffer not found: creating new
2013-02-12 10:08:47 32153 [Note] InnoDB: Doublewrite buffer created
2013-02-12 10:08:47 32153 [Note] InnoDB: 128 rollback segment(s) are active.
2013-02-12 10:08:47 32153 [Warning] InnoDB: Creating foreign key constraint system tables.
2013-02-12 10:08:47 32153 [Note] InnoDB: Foreign key constraint system tables created
2013-02-12 10:08:47 32153 [Note] InnoDB: Creating tablespace and datafile system tables.
2013-02-12 10:08:47 32153 [Note] InnoDB: Tablespace and datafile system tables created.
2013-02-12 10:08:47 32153 [Note] InnoDB: Waiting for purge to start
2013-02-12 10:08:47 32153 [Note] InnoDB: 1.2.10 started; log sequence number 0
2013-02-12 10:08:47 32153 [Warning] /ssd/mysql-5.6.10-linux-glibc2.5-x86_64/bin/mysqld: unknown option '--loose-skip-ndbclus
2013-02-12 10:08:49 32153 [Note] Binlog end
2013-02-12 10:08:49 32153 [Note] InnoDB: FTS optimize thread exiting.
2013-02-12 10:08:49 32153 [Note] InnoDB: Starting shutdown...
===========

However, on the second run (i.e. the actual testrun, whether it be via RQG [which uses MTR] or an actual MTR test) it does a crash recovery:

===========
2013-02-12 02:08:50 32180 [Note] InnoDB: Log scan progressed past the checkpoint lsn 49463
2013-02-12 02:08:50 32180 [Note] InnoDB: Database was not shutdown normally!
2013-02-12 02:08:50 32180 [Note] InnoDB: Starting crash recovery.
2013-02-12 02:08:50 32180 [Note] InnoDB: Reading tablespace information from the .ibd files...
2013-02-12 02:08:50 32180 [Note] InnoDB: Restoring possible half-written data pages
2013-02-12 02:08:50 32180 [Note] InnoDB: from the doublewrite buffer...
InnoDB: Doing recovery: scanned up to log sequence number 1625977
2013-02-12 02:08:50 32180 [Note] InnoDB: Starting an apply batch of log records to the database...
InnoDB: Progress in percent: 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91
 92 93 94 95 96 97 98 99
InnoDB: Apply batch completed
===========

How to repeat:
rm -Rf /tmp/vardir
cd .../mysql-test
perl lib/v1/mysql-test-run.pl --start-and-exit --skip-ndb --vardir=/tmp/vardir --master_port=14247 --mysqld="--innodb_fast_shutdown=2" 1st
vi /tmp/vardir/log/master.err

Suggested fix:
If --mysqld=--innodb_fast_shutdown=2 is specified for MTR, drop it for the first boostrap startup
[15 Feb 2013 18:35] Sveta Smirnova
Thank you for ther report.

Verified as described.

I used command: ./mtr  --mysqld=--innodb_fast_shutdown=2 1st
[9 Apr 2013 6:20] Roel Van de Paar
This continues to create confusion when reviewing test results. 
It would be great if this could be fixed ASAP.
[9 Apr 2013 12:51] Vemund Østgaard
Hi Roel,

Are you using MTR v1 or v2 in this case? Or both?
[25 Nov 2013 16:21] Paul Dubois
Fixed in 5.6.16, 5.7.4.
Changes for test suite. No changelog entry needed.