You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Prior to datanodes update we should add a k8s job, that would set //sys/@config/node_tracker/enable_real_chunk_locations=%true if image version is in a range 23.2 — 24.1. It is required to enable update to 24.2 in the future.
After that we will ask users to update ytop to the version with this fix and run update (full or data nodes only) @ 23.2 or 24.1 version.
N.B. Such update should be done with re-creating all ytsaurus nodes pods, which is default behaviour for full update (which is triggered when master image is updated). Node restart wouldn't happen with updateSelector=MasterOnly, but we consider updateSelector experimental functionality for those who knows what they doing (for example they can restart nodes themselves to have less downtime).
Extra details:
job should run on both cluster create (Initialization state) and update (Updating state) before dnd pods are created/recreated
remote data nodes also should have the same workflow (though if they exist they will require separate update with new ytop version)
The text was updated successfully, but these errors were encountered:
Prior to datanodes update we should add a k8s job, that would set
//sys/@config/node_tracker/enable_real_chunk_locations=%true
if image version is in a range23.2 — 24.1
. It is required to enable update to 24.2 in the future.After that we will ask users to update ytop to the version with this fix and run update (full or data nodes only) @ 23.2 or 24.1 version.
N.B. Such update should be done with re-creating all ytsaurus nodes pods, which is default behaviour for full update (which is triggered when master image is updated). Node restart wouldn't happen with
updateSelector=MasterOnly
, but we considerupdateSelector
experimental functionality for those who knows what they doing (for example they can restart nodes themselves to have less downtime).Extra details:
The text was updated successfully, but these errors were encountered: