Tags: mhartfield/cockroach
Tags
Merge cockroachdb#59633 59633: multiregion: replace TODO and unimplemented with issue numbers r=ajstorm a=otan In the spirit of Milestone planning, replace all remaining multiregion TODOs with issue numbers. Release note: None Co-authored-by: Oliver Tan <otan@cockroachlabs.com>
tracing: revert trace.mode default to legacy We are still seeing memory issues on tpccbench/nodes=6/cpu=16/multi-az which need to be investigated. Turn off background tracing while we do. Touches cockroachdb#58298. We're also reverting an earlier commit as part of this one (d252400). This revert is needed given we've not yet addressed an underlying bug (cockroachdb#59203). Release note: None
Merge pull request cockroachdb#59172 from rytaft/backport20.1-59147 release-20.1: opt: fix bug where ON CONFLICT DO NOTHING ignores some input
Merge pull request cockroachdb#58720 from bdarnell/skippy-20.2
Merge pull request cockroachdb#58876 from bdarnell/skippy-19.2 *: Update gogoproto
Merge pull request cockroachdb#55058 from ajwerner/backport20.1-54876 release-20.1: sql: wait for canceled mutations to be at the head of the queue for reverse
Merge pull request cockroachdb#57677 from ajwerner/backport20.2-57648 release-20.2: sql: add force flag overloads to crdb_internal.unsafe_.* builtins
backupccl: introduce presplit_by_range cluster setting This commit introduces a cluster setting that controls if backup will try to presplit the spans that it imports into range-sized chunks to hand out to its workers. This is a short-term fix intended to by-pass reading meta2, which might be stuck. Release note: None
Merge cockroachdb#57355 57355: kv/concurrency: bump goroutine stall delay r=nvanbenschoten a=nvanbenschoten Fixes cockroachdb#57273. This commit bumps the goroutine stall delay in the TestConcurrencyManagerBasic framework from 1ms to 5ms. In cockroachdb#57273, we saw a flake due to a goroutine being classified as stalled when it was actually waiting for a multi-goroutine chain of events to fire (request exits lockTable, signals lockTable wait-queue, which is noticed by watchForNotifications, which cancels the pushers context). This chain of events plus the 1ms wait in the framework led to a flake under stress roughly every 16k iterations. With this delay bumped to 3ms, I have never reproduced the flake after over 2M iterations. This PR bumps the delay all the way up to 5ms to be extra sure. This increases the runtime of all subtests combined from 1.6s to 3.9s. Co-authored-by: Nathan VanBenschoten <nvanbenschoten@gmail.com>
release-20.1: gss: fix TestComposeGSS Refs cockroachdb#56745 Release note: None
PreviousNext