Redshift Research Project

Blog

2024-01-06

Weather Report

Bit delayed this fortnight; relocating, and New Year.

Problems bringing up ra3 cluster persist.

Region Node Notes
ap-southeast-1 ra3.xlplus Cluster failed to come up.
ap-southeast-3 dc2.large Disk-read-write back to normal (~3s, was ~12.5s).
ap-southeast-3 dc2.large Network back to normal (~4s, was ~9.5s).
ap-southeast-3 ra3.xlplus Disk-read-write back to normal (~3.5s, was ~9.5s).
ap-southeast-3 ra3.xlplus Network back to normal (~2s, was ~6s).
eu-south-1 dc2.large Disk-read-write back to normal (~3s, was ~12.5s).
eu-south-1 dc2.large Network back to normal (~3.5s, was ~9.5s).
eu-south-1 ra3.xlplus Disk-read-write back to normal (~3s, was ~9.5s).
eu-south-1 ra3.xlplus Network back to normal (~1.5s, was ~6.5s).
us-east-1 ra3.xlplus Cluster failed to come up.
us-east-1 ra3.4xlarge Cluster failed to come up.
us-east-1 ra3.16xlarge Cluster failed to come up.
us-west-2 ra3.xlplus Cluster failed to come up.

2024-01-12

Rollback Affecting 16mb SUPER

So, rollbacks in Redshift are not quite what you’d expect.

This is because the version number always increases.

A rollback on the face of it would mean going from version say 5 to version say 6, but then rolling back to version 5.

In Redshift, it’s version 5 to version 6, and then rolling back to version 7, where version 7 has all the new functionality of version 6 removed.

There has been a rollback, where 16mb SUPER was introduced, then rolled back - without warning or notification. Users woke up to find the functionality had gone.

This was all in current track. I do not currently monitor maintenance track, but I need to start doing that.

You can see the rollback on the System Table Tracker page;

https://www.redshiftresearchproject.org

Version Tables Views Functions
1.0.61191 1227 743 2894
1.0.61395 1255 756 2978
1.0.61628 1227 743 2894
1.0.61678 1255 756 2978

Version 1.0.61628 rolled back. Version 1.0.61678 came out shortly after, fixing whatever was wrong. It’s not yet deployed in all regions.

2024-01-15

Web-Site Server Move

I’ve moved the server the RRP site runs on.

The move is now complete.

If there are any hiccups, let me know.

2024-01-16

New STV_RECENTS Issues

Looks like STV_RECENTS has had some work done.

  1. It is now showing -1 for pid, always (this is a bug).
  2. Transaction ID for leader-node queries is now always NULL (so you now can’t find the other queries in the transaction).
  3. It now shows queries which were serviced by the leader node query result cache.
  4. The query text now has actual newlines in, so printing the text, when it originally had newlines, now gives multi-line text.

2024-01-18

Weather Report

Benchmarks were taken 2024-01-15, publication is a bit late (still busy relocating).

ra3 nodes still not coming up properly in some regions. Couldn’t get any started in us-east-1 (in all but two regions, I bring up only ra3.xlplus - us-east-1 is one of those regions, and all ra3 type clusters failed to come up. What happens is they appear to spin up, but you can’t connect to the clusters, they time out.)

region node comments
ap-northeast-2 ra3.xlplus Cluster bring-up failed.
ap-south-1 ra3.xlplus Cluster bring-up failed.
ap-southeast-2 ra3.xlplus Cluster bring-up failed.
ca-central-1 ra3.xlplus Cluster bring-up failed.
eu-south-1 ra3.xlplus Disk-read-write slow, ~7s vs normal ~3.5s.
sa-east-1 ra3.xlplus Cluster bring-up failed.
us-east-1 ra3.xlplus Cluster bring-up failed.
us-east-1 ra3.4xlarge Cluster bring-up failed.
us-east-1 ra3.16xlarge Cluster bring-up failed.
us-west-2 ra3.xlplus Cluster bring-up failed.

Home 3D Друк Blog Bring-Up Times Consultancy Cross-Region Benchmarks Email Forums IRC Mailing Lists Reddit Redshift Price Tracker Redshift Version Tracker Redshift Workbench System Table Tracker The Known Universe Twitter White Papers