[MCOL-1278] occasionally IDB-2031: Blocks are missing. Alter or drop table in progress?` Created: 2018-03-16  Updated: 2020-08-25  Resolved: 2018-11-28

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

Type: Bug Priority: Major
Reporter: Richard Stracke Assignee: Andrew Hutchings (Inactive)
Resolution: Duplicate Votes: 2
Labels: None
Environment:

Debian 8

Issue Links:
Blocks
blocks MCOL-930 Cannot execute queries Closed
Duplicate
duplicates MCOL-1662 WriteEngine bulk methods do not versi... Closed

 Description   

Sometimes for some statements (select or insert) the following error occurs.

Result: `HY000 1815 Internal error: IDB-2031: Blocks are missing. Alter or drop table in progress?`

If the error occurs, the error will repeat for every retry.

Erro vanished, if

calFlushCache()

will be executed.

In error case the following line occured in debug, warning, info, error and critical log

Mar 13 15:53:40 tarcs01 PrimProc[22169]: 40.808502 |0|0|0| C 28 CAL0000: /home/builder/mariadb-columnstore-server/mariadb-columnstore-engine/primitives/primproc/columncommand.cpp error on projectResultRG for oid 350608 lbid 16743424: input rids 876,  output rids 0

execution order:

Mar 13 15:53:38 tarcs01 ExeMgr[22274]: 38.965410 |2350|0|0| D 16 CAL0041: <Statement>
 
 
 
Mar 13 15:53:39 tarcs01 ExeMgr[22274]: 39.793165 |2147485998|0|0| D 16 CAL0041: Start SQL statement: select objectid,columnname from syscolumn  [..]
 
 
 
 
 
Mar 13 15:53:40 tarcs01 PrimProc[22169]: 40.808502 |0|0|0| C 28 CAL0000: /home/builder/mariadb-columnstore-server/mariadb-columnstore-engine/primitives/primproc/columncommand.cpp error on projectResultRG for oid 350608 lbid 16743424: input rids 876,  output rids 0
 
 
 
Mar 13 15:53:40 tarcs01 ExeMgr[22274]: 40.944300 |2147485998|0|0| D 16 CAL0042: End SQL statement

I don't beleive, it is a duplicate from MCOL-631.
portconfigure was succesfully run and local query flag is not set.

infinidb_local_query =0

 Comments   
Comment by Andrew Hutchings (Inactive) [ 2018-04-17 ]

Ah! So my guess is the DDL doesn't clear PrimProcs cache properly (likely on dict) so it is expecting blocks from the old table instead of the replacement one. I'll assign it to 1.1.5 but we will need to figure out a way to reproduce it.

Comment by Stepan Patryshev (Inactive) [ 2018-08-20 ]

Hi Richard, can you please share with me a sample data file?

Comment by Richard Stracke [ 2018-08-22 ]

It could be the same issue like MCOL 930, both issues can be "solved" with

select calFlushCache();

Comment by Andrew Hutchings (Inactive) [ 2018-11-28 ]

Duplicate of MCOL-1662 where I am working on a fix for this.

Generated at Thu Feb 08 02:27:33 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.