I made fundamental changes to cluster bring up, to try to work around the now long-standing problems with clusters reporting availability but in fact not being able to accept connections.
This has helped, and I now have the large ra3 nodes coming up successfully, but still in one or two regions, one or two ra3 clusters did not become available in a timely manner and after a couple of full retries (delete cluster, make new cluster).
| region | node | comments |
|---|---|---|
| ap-northeast-1 | dc2.large | Disk read-write slow. |
| ap-northeast-2 | dc2.large | Disk read-write slow, networking slow. |
| ap-northeast-3 | dc2.large | Disk read-write slow, networking slow. |
| ap-south-1 | dc2.large | Disk read-write slow, networking slow. |
| ap-southeast-1 | dc2.large | Disk read-write slow, networking slow. |
| ap-southeast-2 | dc2.large | Disk read-write slow, networking slow. |
| eu-central-1 | ra3.xlplus | Cluster failed to start. |
| eu-west-1 | ra3.xlplus | Cluster failed to start. |
| us-west-1 | dc2.large | Disk read-write slow, networking slow. |
| us-west-2 | dc2.large | Networking slow. |
https://www.redshiftresearchproject.org/cross_region_benchmarks/index.html
(Thinking to take a day or two off and build some code to make ongoing gnuplots of all the data :-)
It finally happened :-)
https://www.redshiftresearchproject.org/cross_region_benchmarks/index.html
The version tracker is now also tracking maintenance track. Previously, it was current track only.
Current maintenance release is from 27th Jan - and today is 19th April!
But we’ve had now no new release on current since 1st March, a big long gap.
https://www.redshiftresearchproject.org/redshift_version_tracker/index.html
After about six or seven weeks, a new Redshift version is out, on the current track.
Currently in four regions, ap-south-1, eu-central-1, us-east-2, us-west-2.
https://www.redshiftresearchproject.org/redshift_version_tracker/index.html
Maintenance track hasn’t changed.
Cluster admin are reporting 5x query slow-downs, and are diving for safety to the maintenance track (and I have to say, I was shocked anyone is running current track - Redshift just isn’t reliable enough for that).
One admin reporting problems say it was bad in the morning, but has seems better now.
I’ve dumped the system tables for the new release, and they’ve not changed; same number of functions, tables and views, with no changes.
Well, I have some interesting and completely unexpected news.
First, a couple more regions have moved up to the new 1.0.66954 version on current. Fine and dandy.
But, to my complete surprise, FOUR REGIONS ONLY have moved up on trailing, and - get this - to version 1.0.66960 (!!!)
I had thought all regions moved together on trailing, and to a version which had already been in current.
https://www.redshiftresearchproject.org/redshift_version_tracker/index.html
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