release: v19.2.0-beta.20190826

[dt]

Candidate SHA: bd27eb3
Deployment status: david-release-v1920-beta20190826
Nightly Suite

Release process checklist

Prep date: WIP fill in date to roll out: usually Monday, a week before the release WIP

• Pick a SHA, fill in Candidate SHA above, notify #release-process of SHA.

• Tag the provisional SHA

• Publish provisional binaries

• Check binaries

• Deploy to test clusters

• Verify node crash reports

• Start release qualification suite: build

• Start nightly suite: build

• fill in Deployment status above with clusters and Nightly Suite with the link to Nightly TeamCity Job

• Keep an eye on clusters until release date. Do not proceed below until the release date.

Release date: WIP fill in date: usually Monday WIP

• Check cluster status

• Tag release

• Bless provisional binaries

• Update docs

• External communications for release

[dt]

Failed roachtest nightlies:

• acceptance/bank/cluster-recovery @nvanbenschoten based on roachtest: acceptance/bank/cluster-recovery failed #38785
• acceptance/bank/node-restart @andreimatei based on roachtest: acceptance/bank/node-restart failed #35326
• acceptance/build-analyze @dt looks like apt-get flaked, subsequent run passed.
• acceptance/cli/node-status
• acceptance/decommission this passed in the followup run.
• backup2TB/n10cpu4 roachtest: backup2TB/n10cpu4 failed #38794 @dt importccl: avoid double ProducerDone call #39992
• cancel/tpcc/distsql/w=10,nodes=3 @jordanlewis based on roachtest: cancel/tpcc/distsql/w=10,nodes=3 failed #38417.
• import/tpch/nodes=4 @dt: We'll look into it but it isn't blocking: this has been timing out for awhile so this isn't a regression compared to the alpha.
• overload/tpcc_olap/nodes=3/cpu=16/w=100/c=256 Looks like roachtest: overload/tpcc_olap/nodes=3/cpu=16/w=100/c=256 failed #39684 so @ajwerner, these have been determined to OOM on GCP because for whatever reason we have half the RAM for same number of cores as the AWS test, it also was exposed to roachtest: overload/tpcc_olap/nodes=3/cpu=16/w=100/c=256 failed #39684. It's not a regression.

Failed RSG: bulk-signoff from @mjibson yes, move along. rsg/sqlsmith should never block releases.

• TestRandomSyntaxFunctions
• TestRandomSyntaxGeneration
• TestRandomSyntaxSQLSmith

[dt]

The qualification cluster encountered a crash on node 5: storage/replica_raft.go:668 [n5,s5,r3157/5:{-}] setting non-zero HardState.Commit on uninitialized replica [n5,s5,r3157/5:{-}]. HS={Term:7 Vote:0 Commit:66957 XXX_unrecognized:[]}

It looks like this might be #39660 ? I've left node 5 as-in in case anyone on core wants to gather any data from it.

[tbg]

^- #40213 has a repro and first insights, I think we can figure this out in a day or two

[dt]

This beta has been aborted while we pursue a fix for #40213, which should then be in the next beta.