Understanding how history_node table works

Community Support
,
1

Hi,

My history_node table is huge (~16GB) and I would like to understand why and when exactly is this table updated. My first question is related to the Archival page in your documentation. You say there that Archival copies “closed Workflow Execution Event Histories and Visibility records from Temporal Service persistence to blob storage”. Is history_node table part of the Temporal Service persistence? Or it is part of the blob storage?

Thanks,

Tiberiu

2

Is history_node table part of the Temporal Service persistence? Or it is part of the blob storage?

its server persistence, it stores batches of history events for your executions

My history_node table is huge (~16GB) and I would like to understand why and when exactly is this table updated.

each time new events for execution are persisted. what can make it large is large event histories (in terms of both event count and event history size for example large payloads, lots of executions that do ContinueAsNew as well

closed execution event histories are cleaned up from this table on reaching namespace retention
open (running) executions are never considered for cleanup until they close, and hit ns retention

if you have longer namespace retention time configured, event history batches for closed execs can sit in this table longer
table is not cleaned by archival, archival copies even histories to your blob store configured, cleaning done by server on closed hitting ns retention