[MCOL-778] Create table failed due to Lost connection to WriteEngineServer on pm0 Created: 2017-06-20  Updated: 2017-08-09  Resolved: 2017-08-09

Status: Closed
Project: MariaDB ColumnStore
Component/s: installation
Affects Version/s: 1.0.9
Fix Version/s: Icebox

Type: Bug Priority: Major
Reporter: dic (Inactive) Assignee: Unassigned
Resolution: Cannot Reproduce Votes: 0
Labels: None
Environment:

CentOS 7.x

Attachments: File columnstoreSupportReport.mcs-1.tar.gz    

 Description   

Error when create a table:
MariaDB [(none)]> create database mcs;
Query OK, 1 row affected (0.01 sec)

MariaDB [(none)]> use mcs;
Database changed
MariaDB [mcs]>
MariaDB [mcs]> create table idbtest(col1 int, col2 int) engine=columnstore;
ERROR 1815 (HY000): Internal error: CAL0009: (2)Create table failed due to Lost connection to WriteEngineServer on pm0

System Info:

  1. mcsadmin getsysteminfo
    getsysteminfo Tue Jun 20 14:31:07 2017

System mcs-1

System and Module statuses

Component Status Last Status Change
------------ -------------------------- ------------------------
System ACTIVE Tue Jun 20 14:26:48 2017

Module pm1 ACTIVE Tue Jun 20 14:26:43 2017

MariaDB ColumnStore Process statuses

Process Module Status Last Status Change Process ID
------------------ ------ --------------- ------------------------ ----------
ProcessMonitor pm1 ACTIVE Tue Jun 20 11:03:51 2017 15603
ProcessManager pm1 ACTIVE Tue Jun 20 11:03:57 2017 15687
DBRMControllerNode pm1 ACTIVE Tue Jun 20 14:26:20 2017 4256
ServerMonitor pm1 ACTIVE Tue Jun 20 14:26:22 2017 4273
DBRMWorkerNode pm1 ACTIVE Tue Jun 20 14:26:22 2017 4294
DecomSvr pm1 ACTIVE Tue Jun 20 14:26:26 2017 4350
PrimProc pm1 ACTIVE Tue Jun 20 14:26:28 2017 4368
ExeMgr pm1 ACTIVE Tue Jun 20 14:26:32 2017 4418
WriteEngineServer pm1 ACTIVE Tue Jun 20 14:26:36 2017 4487
DDLProc pm1 ACTIVE Tue Jun 20 14:26:40 2017 4521
DMLProc pm1 ACTIVE Tue Jun 20 14:26:45 2017 4553
mysqld pm1 ACTIVE Tue Jun 20 14:26:15 2017 4144

Active Alarm Counts: Critical = 0, Major = 0, Minor = 0, Warning = 0, Info = 0

 Comments   
Comment by Andrew Hutchings (Inactive) [ 2017-06-20 ]

Can you please run the MariaDB ColumnStore support tool and attach the results here?

https://mariadb.com/kb/en/mariadb/system-troubleshooting-mariadb-columnstore/#mariadb-columnstore-support-tool

Comment by dic (Inactive) [ 2017-06-20 ]

I updated results columnstoreSupportReport.mcs-1.tar.gz after run the MariaDB ColumnStore support tool

Comment by David Hill (Inactive) [ 2017-06-20 ]

This issue is the the System Catalog is created and there is no useable extent map in the DBRM files.

The system catalog gets created at the end of postConfigure script, which is run as part of the install. So if this script error out or didnt complete, then that would explain the missing system catalog

Here is how to create the system catalog manually

on pm1:

  1. /usr/local/mariadb/columnstore/bin/dbbuilder 7
    If this runs successfully, then try the create table
Comment by dic (Inactive) [ 2017-06-21 ]

It can not completed:

  1. # /usr/local/mariadb/columnstore/bin/dbbuilder 7
    Build system catalog was not completed. System catalog appears to exist. It will remain intact for reuse. The database is not recreated.
Comment by David Hill (Inactive) [ 2017-06-21 ]

Ok, so this means you have a partial System Catalog setup.. Still points to some install issue.

So you have 2 options:

1. Do a complete uninstall and reinstall and make sure it completes postConfigure
1. erase package on all servers
2. # rm -rf /usr/local/mariadb
3. install and run postConfigure from pm1

2. Keep the install, but delete the current DB and rebuild the system Catalog

Procedure to get a system to a fresh DB install state
1. ma shutdownsystem y // from pm1
2. rm -rf /usr/local/mariadb/columnstore/data*/000.dir // run on all pm nodes
3. rm -rf /usr/local/mariadb/columnstore/data1/systemFiles/dbrm/* // run on pm1
4. on all um or combo pm front-end nodes
cd /usr/local/mariadb/columnstore/mysql/db
delete all directories except
calpontsys, infinidb_querystats, infinidb_vtable, mysql, performance_schema, test
5. /usr/local/mariadb/columnstore/bin/clearShm // run on all nodes
6. ma startsystem // from pm1, if multi-node, provide root password
7. /usr/local/mariadb/columnstore/bin/dbbuilder 7

Comment by David Thompson (Inactive) [ 2017-07-12 ]

nguyen - were you able to resolve this issue with the steps provided by david above?

Generated at Thu Feb 08 02:23:45 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.