Details
-
Bug
-
Status: Closed (View Workflow)
-
Major
-
Resolution: Fixed
-
1.5.2, 5.5.1
-
None
-
CentOS 7.8.2003 / packages from bb-10.5-cs 19d09e49912, tarbuildnum 33657 ; VirtualBox 6.1, Debian Stretch host, guests installed from the official .iso images, run with default VirtualBox settings (only memory and disk size adjusted)
Description
On a clean CentOS VM with RPMs downloaded on the local disk from a buildbot build, the following steps are performed:
[elenst@localhost rpms]$ sudo yum install MariaDB-server-10.5.4-1.el7.centos.x86_64.rpm MariaDB-client-10.5.4-1.el7.centos.x86_64.rpm MariaDB-shared-10.5.4-1.el7.centos.x86_64.rpm MariaDB-compat-10.5.4-1.el7.centos.x86_64.rpm MariaDB-common-10.5.4-1.el7.centos.x86_64.rpm galera-4-26.4.4-1.rhel7.el7.centos.x86_64.rpm MariaDB-columnstore-engine-10.5.4-1.el7.centos.x86_64.rpm
|
[elenst@localhost rpms]$ sudo systemctl start mariadb
|
[elenst@localhost rpms]$ sudo mysql -e "create database if not exists test; use test; create table t1 (a int) engine=Columnstore; insert into t1 values (1); select * from t1"
|
The latter results in
ERROR 1815 (HY000) at line 1: Internal error: CAL0009: Error while calling getSysCatDBRoot
|
$ cat columnstore.centos7.bad/crit.log
|
Jun 19 12:41:16 localhost writeengine[4643]: 16.661628 |0|0|0| C 19 CAL0060: dbbuilder system catalog error: Creating TableName column OID: 1001 a BRM Allocate extent error. [BRM error status: network error]
|
It happened every of several times when I tried – each time from the freshly restored snapshot: install, start, run the command => error. However, there still can be some race condition involved which I am not aware of.
This, on the other hand, seems to always end well:
[elenst@localhost rpms]$ sudo yum install MariaDB-server-10.5.4-1.el7.centos.x86_64.rpm MariaDB-client-10.5.4-1.el7.centos.x86_64.rpm MariaDB-shared-10.5.4-1.el7.centos.x86_64.rpm MariaDB-compat-10.5.4-1.el7.centos.x86_64.rpm MariaDB-common-10.5.4-1.el7.centos.x86_64.rpm galera-4-26.4.4-1.rhel7.el7.centos.x86_64.rpm MariaDB-columnstore-engine-10.5.4-1.el7.centos.x86_64.rpm
|
[elenst@localhost rpms]$ sudo systemctl start mariadb
|
[elenst@localhost rpms]$ sudo mysql -e "create database if not exists test; use test; create table t1 (a int) engine=Columnstore; insert into t1 values (1); select count(*) from t1"
|
[elenst@localhost rpms]$ sudo mysql -e "select * from test.t1"
|
and results in
+----------+
|
| count(*) |
|
+----------+
|
| 1 |
|
+----------+
|
|
|
+------+
|
| a |
|
+------+
|
| 1 |
|
+------+
|
The only difference is the SQL commands.
The attached archives contain logs from /var/log/mariadb/columnstore, from the good and bad attempt. Additionally both include output files with the unabridged output of the commands above.
openSUSE 15.1 seems to produce similar results, but the number of attempts was too low to make any conclusions.
On Ubuntu Xenial, the "bad" sequence appears to work in general. I did have issues while executing it, but they are likely to be unrelated.
I have a CentOS VM image stored after a "bad" attempt and can provide additional information from there upon request.