XtraDB-cluster-56 crashed on database update

[Sergey]

Hello,

I found some strange bug, Percona-XtraDB-Cluster-server-56 crashed and lost some data in database but Percona-XtraDB-Cluster-server-55 works well.

Current setup - db_server-1:
- FreeBSD 9.1-RELEASE-p22 amd64
- percona55-server-5.5.38.35.2 with database size 52GB
- rt42-4.2.1_1 (request tracker from bestpractial.com)

New setup with 3 x Nodes:
- lates CentOS release 6.6 (Final) 2.6.32-504.el6.x86_64
- Percona-XtraDB-Cluster-server-56.x86_64 1:5.6.20-25.7.888.el6 with all dependencies from percona repo
- rt-4.2.9

Steps:
1. dump database with innobackupex on db_server-1
2. copy dump to Node1
3. restore dump with “innobackupex --copy-back” and chown mysql:mysql /home/mysql
4. bootstrap first Node1 server
5. mysql-upgrade for upgrade database from 55 to 56
6. cd src/rt-4.2.9; ./configure ; make install; make upgrade-database (for upgrade RT database from 4.2.1 to 4.2.9)
7. crashed on 4.2.2 (corefile core.35841 in archive )
8. bootstrap Node1 server again
9. run “make upgrade-database” again, from 4.2.2 to 4.2.9
7. crashed on 4.2.6 (corefile core.36765 if need I can provide this file too)
8. bootstrap Node1 server again
9. run “make upgrade-database” again, from 4.2.6 to 4.2.9
7. not crashed
10. found what in table Users(example), only 90000 records, when on db_server-1 - 96979, so on step 7 we lost some data in database, what is not good

Same steps(1-4,6) on same server but with Percona-XtraDB-Cluster-server-55.x86_64 1:5.5.39-25.11.824.el6 without crash, and all works well.

Screenshot with crash: https://www.dropbox.com/s/m9ii82mbuv8rabf/Screenshot%202014-11-10%2013.22.02.png?dl=0
core.35841: https://www.dropbox.com/s/b8kjcxrfoj05g26/core.35841.tgz?dl=0
my.cnf: https://www.dropbox.com/s/qrs6xvr2e12yqgq/my.cnf?dl=0
error.log: https://www.dropbox.com/s/7ae00q4vsumelx9/node1.err?dl=0

[Shahriyar Rzayev (rzayev-sehriyar)]

Percona now uses JIRA for bug reports so this bug report is migrated to: https://jira.percona.com/browse/PXC-1764