wsrep/xtrabackup-v2 fails when having a dedicated binary logfile directory

MySQL & MariaDB Percona XtraDB Cluster 5.x
1

I’ve already setup the donor and now want to start the second node.

rpm -qa | grep -i percona
Percona-XtraDB-Cluster-client-56-5.6.21-25.8.938.el6.x86_64
percona-xtrabackup-2.2.7-5050.el6.x86_64
Percona-XtraDB-Cluster-shared-56-5.6.21-25.8.938.el6.x86_64
Percona-XtraDB-Cluster-server-56-5.6.21-25.8.938.el6.x86_64
percona-release-0.1-3.noarch
Percona-XtraDB-Cluster-galera-3-3.8-1.3390.rhel6.x86_64

sst-method is set to ‘xtrabackup-v2’.
With a dedicated binlog directory ‘/db03/mysql56-01/binlogs’, wsrep fails with the following messages:

WSREP_SST: [INFO] Cleaning the existing datadir and innodb-data/log directories (20141219 09:03:29.673)
removed /db02/mysql56-01/innodata/undo001' removed /db02/mysql56-01/innodata/ibdata2’
removed /db02/mysql56-01/innodata/auto.cnf' removed /db02/mysql56-01/innodata/ibdata1’
removed /db04/mysql56-01/innologs/ib_logfile1' removed /db04/mysql56-01/innologs/ib_logfile0’
dirname: extra operand /db03/mysql56-01/binlogs/mysql56-01-bin' Try dirname --help’ for more information.
WSREP_SST: [ERROR] Cleanup after exit with status:1 (20141219 09:03:29.770)
2014-12-19 09:03:29 17264 [ERROR] WSREP: Process completed with error: wsrep_sst_xtrabackup-v2 --role ‘joiner’ --address ‘192.168.137.171’ --auth ‘root:mysql’ --datadir ‘/db02/mysql56-01/innodata/’ --defaults-file ‘/db05/mysql56-01/admin/startup/my.cnf’ --parent ‘17264’ ‘’ : 1 (Operation not permitted)
2014-12-19 09:03:29 17264 [ERROR] WSREP: Failed to read uuid:seqno from joiner script.
2014-12-19 09:03:29 17264 [ERROR] WSREP: SST failed: 1 (Operation not permitted)
2014-12-19 09:03:29 17264 [ERROR] Aborting

The ‘dirname’ command is somehow used with the wrong amount of arguments, i guess.

The whole err-log is attached.

Kind regards,
Wolfsrudel

error-log.zip (4.64 KB)

2

It seems like a pretty non-standard directory structure layout, hence the problems. Is there one or more MySQL/PXC instances on the same box? I can see dirs /db02, /db04, /db05… Can you share the configs?

3

The configuration is based on MOCA ([url]http://mysql-dba-journey.blogspot.de/2013/04/best-practices-for-installing-mysql-moca.html[/url]). Software, data and logfiles are seperated on different disks. Two instances are running on the box: one one port 3306, the other on port 3307.

4

Configs attached. “mysql56-01” and “mysql56-02” are the names of the instances.

configs.zip (9.29 KB)

5

Your configs are very confusing. You use different main directories for the same instance parts, like /db02/ and /db05/, etc. :slight_smile:

Anyway, I can’t see the
[url]Index of wsrep system variables
set in wsrep.cnf, try using it.

Also this may be helpful:
[url]Percona XtraDB Cluster

6

Thanks, i’ll give it a try.

/db02 and /db05 are not different instances. /db02 is for datafiles (from instance 1 and 2), /db03 for binlogs, etc. At Oracle it is called “OFA”. You put datafile, logfiles and software installtion on different hdds for a better performance. So /dbx are hdds, not instances. :wink:

I’ve installed the cluster on two new testing hosts with default directory structure (everything into /var/lib/mysql) and it’s working. I guess splitting the files into different directories is not supported 'til now. :frowning: