[MCOL-888] WriteEngineServ segfault .. in libstdc++.so.6.0 Created: 2017-08-22 Updated: 2021-01-16 Resolved: 2021-01-16 |
|
| Status: | Closed |
| Project: | MariaDB ColumnStore |
| Component/s: | writeengine |
| Affects Version/s: | 1.0.10 |
| Fix Version/s: | N/A |
| Type: | Bug | Priority: | Critical |
| Reporter: | Nivesh | Assignee: | Unassigned |
| Resolution: | Won't Fix | Votes: | 0 |
| Labels: | None | ||
| Environment: |
Centos 7 |
||
| Attachments: |
|
| Description |
|
We have moved our environment to centos 7 we get a writeengin crash I've added the crit log which has errors as below. Aug 22 09:46:24 controllernode[14284]: 24.230055 |0|0|0| C 29 CAL0000: ExtentMap::getLocalHWM(): invalid OID requested: -1839475712 I have only attach the crit log as it is the only log I have sanitized . |
| Comments |
| Comment by David Hill (Inactive) [ 2017-08-22 ] |
|
Not a happy system. It looks like the Extent Map entries aren't matching the database, which is causing the issue. Extent Map is in the DBRM files in pm1 data1 directory. 1 question, how did you go about moving from the old env to this new Centos 7 environment. What was the procedure you followed and what was the old Env, Centos 6, for example. |
| Comment by Nivesh [ 2017-08-23 ] |
|
Hi David we have moved every database and table manually row by row from the centos 6 infinidb system to the new centos 7 mairadb cs system here are some more errors I got while testing only user load on this environment. with no table loads or ddl happening, ERROR 1815 (HY000) at line 1412: Internal error: An unexpected condition within the query caused an internal processing error within InfiniDB. Please check the log files for more details. Additional Information: error in BatchPrimitiveProces ==> /var/log/mariadb/columnstore/info.log <== ==> /var/log/mariadb/columnstore/debug.log <== ==> /var/log/mariadb/columnstore/crit.log <== ==> /var/log/mariadb/columnstore/err.log <== ==> /var/log/mariadb/columnstore/warning.log <== ==> /var/log/mariadb/columnstore/info.log <== ==> /var/log/mariadb/columnstore/debug.log <== ==> /var/log/mariadb/columnstore/info.log <== ==> /var/log/mariadb/columnstore/debug.log <== ==> /var/log/mariadb/columnstore/crit.log <== ==> /var/log/mariadb/columnstore/err.log <== ==> /var/log/mariadb/columnstore/warning.log <== ==> /var/log/mariadb/columnstore/info.log <== ==> /var/log/mariadb/columnstore/debug.log <== ==> /var/log/mariadb/columnstore/crit.log <== ==> /var/log/mariadb/columnstore/err.log <== ==> /var/log/mariadb/columnstore/warning.log <== ==> /var/log/mariadb/columnstore/info.log <== ==> /var/log/mariadb/columnstore/debug.log <== ==> /var/log/mariadb/columnstore/info.log <== ==> /var/log/mariadb/columnstore/debug.log <== ==> /var/log/mariadb/columnstore/crit.log <== ==> /var/log/mariadb/columnstore/err.log <== ==> /var/log/mariadb/columnstore/warning.log <== ==> /var/log/mariadb/columnstore/info.log <== ==> /var/log/mariadb/columnstore/debug.log <== ==> /var/log/mariadb/columnstore/crit.log <== ==> /var/log/mariadb/columnstore/err.log <== ==> /var/log/mariadb/columnstore/warning.log <== |
| Comment by Daniel Lee (Inactive) [ 2017-08-23 ] |
|
Few questions: 1) What was the old configuration (How many UMs and PMs)? Thx |
| Comment by Nivesh [ 2017-08-24 ] |
|
Hi David. 1) The configuration from old -> new 2) Method used to migrate. I'm just pasting the script below. 3) 16 TB of data was moved. 4) The is a production environment. I have discovered that only certain tables have this issue. |
| Comment by Nivesh [ 2017-08-27 ] |
|
I have been able to reproduce the error here is the CREATE TABLE `test3` ( CREATE TABLE `test3_stg` ( insert into test3_stg values (1,'Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.!@#$%()?><:{}[];,./|= --repeat above insert script 200000X insert into test3 select * from test3_stg; --Crash Writeengine and primproc with overused memory or a faster way is to kill the primproc and write engine process on any node. |
| Comment by Nivesh [ 2017-08-28 ] |
|
FYI we run the database on ext4 with no journaling or caching |
| Comment by Andrew Hutchings (Inactive) [ 2017-08-29 ] |
|
For the INSERT...SELECT are you doing it as part of a transaction or with autocommit? If it is part of a transaction can you please try as an autocommit? |
| Comment by Nivesh [ 2017-08-30 ] |
|
The insert select is run as an "autocommit" not in a transaction. and the mysql flag for cpimport on insert into is set. when monitoring the insert...select I see a cpimport process running. This is only the easiest way to recreate this issue. |
| Comment by Andrew Hutchings (Inactive) [ 2017-08-30 ] |
|
Ah, I see. Yes, the primproc crash would happen depending on the amount of system RAM. Not quite sure why WriteEngine would crash at that point, it shouldn't be used here. But it in any case there appears to be corruption of the column files which we should look into. |
| Comment by Nivesh [ 2017-09-04 ] |
|
Hi do you have any idea of what is causing this? Will this be fixed soon. we have had to rollback to infinidb and we have the same issues with primproc/writeengine |
| Comment by Andrew Hutchings (Inactive) [ 2017-09-04 ] |
|
Hi, There is no update at this time, we will try to reproduce it soon. Thanks for letting us know about the InfiniDB testing. |
| Comment by Nivesh [ 2017-11-02 ] |
|
Hi Has anyone been able to replicate this issue? |
| Comment by Roman [ 2020-03-21 ] |
|
This issue could be potentially caused by the default Linux memory allocator used before 1.2.3. |