CAL0000 Error: Invalid Range from HWM for lbid 4194247

[aeae81]

CAL0000: Invalid Range from HWM for lbid 4194247, range size should be <= blocksReadAhead: HWM 2, dbroot 1, highfbo 3071, lowfbo 2560, blocksReadAhead 512, range size -2557

Hello Group,

I'm writing to request assistance with a CAL0000 error message I'm encountering. The error details are as follows:

• Error: Invalid Range from HWM for lbid 4194247
• Issue: The range size (-2557) is greater than the blocksReadAhead value (512).
• Relevant Values: HWM 2, dbroot 1, highfbo 3071, lowfbo 2560

I'm unsure how to proceed. I'm wondering if reinstalling the software or checking for disk issues might resolve this problem.

Thank you for your time and assistance.

Best regards, Andreas

[Roman Nozdrin]

Hey Andreas,

There is in-memory map of where the actual Columnstore data is in the cluster(even single node is a cluster). It is called ExtentMap.
The message like this usually tells that ExtentMap data is inconsistent.
HighWaterMark is the number of blocks currently used in a group of columnar values(called Extent) from the perspective of Extent Map for a specific column. HWM values is 2. On the contrary you can see `lowfbo 2560` that tells that the file with the columnar values has 2560 blocks used. This inconsistency might be caused by a problematic transaction or killed WriteEngineServer.
Which version you are at? Previously `mcs-writeengineserver` systemd unit depended on PrimProc/ExeMgr that might crash and causes systemd to force restart WE killing DML operation.

Regards,

Roman

[Antonis Panagi]

I stopped using MariaDb. Migrated to postgres :)

--
You received this message because you are subscribed to the Google Groups "MariaDB ColumnStore" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mariadb-columns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mariadb-columnstore/ebf04c04-321a-437c-ae3a-6214f4df587fn%40googlegroups.com.

[Andreas Eschbacher]

Hi Antonis, thank you for your honest reply. To be honest we are currently testing cloudberrydb to replace mariadbcs which is multi parallel postgres. 😄

[Roman Nozdrin]

This is true. With the parallel query execution feature small to moderate datasets doesn't need columnar engines nowadays.

[Roman Nozdrin]

Ah, another incarnation of Greenplum. Looks promising but we will see. Might not have enough steam for a long run. Now telling outside of MCS context.

Regards,

Roman