Bug #109326 MaxDiskDataLatency causes datanode crash

• Submitted: 9 Dec 2022 11:35
• Modified: 13 Dec 2022 8:30
• Reporter: Hendrik Woltersdorf
• Status: Not a Bug
• Category: MySQL Cluster: Cluster (NDB) storage engine
• Severity: S3 (Non-critical)
• Version: 8.0.31
• OS: Any
• Assigned to: MySQL Verification Team
• CPU Architecture: Any

[9 Dec 2022 11:35] Hendrik Woltersdorf

Description:
I tried to use the configuration parameter MaxDiskDataLatency=8000.
This caused the first datanode, where I tried ist, to crash right after finishing its restart.

How to repeat:
use MaxDiskDataLatency=8000

[9 Dec 2022 11:36] Hendrik Woltersdorf

logs and configuration from ndb_error_reporter

Attachment: ndb_error_report_20221209122220.zip (application/x-zip-compressed, text), 842.08 KiB.

[12 Dec 2022 13:11] MySQL Verification Team

Hi Mr. Woltersdorf,

The issue that you are facing is properly documented in our Reference Manual.

Since you have put that variable at its maximum value, it means that your setup is too slow for NDB. 

You should consider either using several different HDDs for each of the schemas or you need to use SSD instead of the HDD.

It does not seem likely that your filesystem is over-fragmented, but you can check that as well.

Simply, your system seems to be  too slow for an NDB installation .......

Not a bug.

[13 Dec 2022 8:30] Hendrik Woltersdorf

The documentation says:
"When this limit is reached, NDB begins to abort transactions in order to decrease pressure on the Disk Data I/O subsystem."

IMHO aborting transactions is something very different then a forced node shutdown.

But the virtual disks of this system ARE slow.