[MCOL-1056] Internal error: InetStreamSocket::readToMagic: Remote is closed Created: 2017-11-29  Updated: 2019-07-10  Resolved: 2019-07-10

Status: Closed
Project: MariaDB ColumnStore
Component/s: ?
Affects Version/s: 1.1.2
Fix Version/s: Icebox

Type: Bug Priority: Major
Reporter: hiller1 Assignee: Unassigned
Resolution: Incomplete Votes: 0
Labels: None
Environment:

MariaDB-log Columnstore 1.1.2-1

1um,2pm

Attachments: JPEG File crash.jpg    

 Description   

MariaDB [(none)]> SELECT w.withdraw_apply_no, w.state, ir.* FROM holmes_analysis.risk_assess_flow_index_result ir LEFT JOIN nirvana.withdraw_apply w on w.withdraw_risk_assess_apply_id = ir.risk_asess_flow_state_id where ir.channel_code = 'qianzhan' and ir.risk_assess_flow_id = 103 and ir.`key1` REGEXP 'TSHATFR001|TSHATFR002|TSHATFR003|TSHATFR004|TSHATFR005' and ir.create_time >= '2017-11-03';
ERROR 1815 (HY000): Internal error: InetStreamSocket::readToMagic: Remote is closed
MariaDB [(none)]>
MariaDB [(none)]>
MariaDB [(none)]> SELECT w.withdraw_apply_no, w.state, ir.id FROM holmes_analysis.risk_assess_flow_index_result ir LEFT JOIN nirvana.withdraw_apply w on w.withdraw_risk_assess_apply_id = ir.risk_asess_flow_state_id where ir.channel_code = 'qianzhan' and ir.risk_assess_flow_id = 103 and ir.`key1` REGEXP 'TSHATFR001|TSHATFR002|TSHATFR003|TSHATFR004|TSHATFR005' and ir.create_time >= '2017-11-03' limit 1;
----------------------------------

withdraw_apply_no state id

----------------------------------

106171108303012930 4 258820079

----------------------------------
1 row in set (6.93 sec)

select ir.* is crash!!!

select ir.id is no problem.

 Comments   
Comment by David Thompson (Inactive) [ 2017-11-29 ]

If i was to take a guess adding ir.* results in a too large intermediate result set causing out of memory error. You should be able to determine the cause of crash from the logs in /var/log/mariadb/columnstore or send us over a columnstoreSupportReport.

Comment by hiller1 [ 2017-11-29 ]

thanks!!!

Comment by Muhammad Abbas [ 2018-04-18 ]

Same error is appearing at my end , what i have seen is that all the data is trying to get into RAM. I have 64B of RAM on my VM. So when i see the result with free -h on VM i can see RAM getting drop from 62GB to 200M and then it crashes. Can you please provide a solution that it does not write over RAM all the results and use disk.

I have also made below changes as well:

/usr/local/mariadb/columnstore/bin/configxml.sh setconfig SystemConfig SwapAction none
/usr/local/mariadb/columnstore/bin/configxml.sh setconfig SystemConfig MemoryCheckPercent 80
/usr/local/mariadb/columnstore/bin/configxml.sh setconfig SystemConfig hdfsRdwrBufferMaxSize 2G

/usr/local/mariadb/columnstore/bin/configxml.sh setconfig SystemConfig AllowDiskBasedJoin Y
/usr/local/mariadb/columnstore/bin/configxml.sh setconfig HashJoin AllowDiskBasedJoin Y

set global infinidb_um_mem_limit=2500;

Comment by Andrew Hutchings (Inactive) [ 2019-07-10 ]

Probably fixed by now

Generated at Thu Feb 08 02:25:51 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.