roachtest: ycsb/B/nodes=3/cpu=32/mvcc-range-keys=global failed

[cockroach-teamcity]

roachtest.ycsb/B/nodes=3/cpu=32/mvcc-range-keys=global failed with artifacts on release-22.2 @ 8afb672bdbe2510330208e5b22f76494a5b3b3d9:

          | main.execCmdEx
          |     main/pkg/cmd/roachtest/cluster.go:382
          | main.execCmd
          |     main/pkg/cmd/roachtest/cluster.go:270
          | main.(*clusterImpl).RunE
          |     main/pkg/cmd/roachtest/cluster.go:1997
          | main.(*clusterImpl).Run
          |     main/pkg/cmd/roachtest/cluster.go:1975
          | github.com/cockroachdb/cockroach/pkg/cmd/roachtest/tests.registerYCSB.func1.1
          |     github.com/cockroachdb/cockroach/pkg/cmd/roachtest/tests/ycsb.go:89
          | main.(*monitorImpl).Go.func1
          |     main/pkg/cmd/roachtest/monitor.go:105
          | golang.org/x/sync/errgroup.(*Group).Go.func1
          |     golang.org/x/sync/errgroup/external/org_golang_x_sync/errgroup/errgroup.go:74
          | runtime.goexit
          |     GOROOT/src/runtime/asm_amd64.s:1594
        Wraps: (5) one or more parallel execution failure
        Error types: (1) *withstack.withStack (2) *errutil.withPrefix (3) *cluster.WithCommandDetails (4) *withstack.withStack (5) *errutil.leafError

    monitor.go:127,ycsb.go:92,ycsb.go:130,test_runner.go:930: monitor failure: monitor task failed: t.Fatal() was called
        (1) attached stack trace
          -- stack trace:
          | main.(*monitorImpl).WaitE
          |     main/pkg/cmd/roachtest/monitor.go:115
          | main.(*monitorImpl).Wait
          |     main/pkg/cmd/roachtest/monitor.go:123
          | github.com/cockroachdb/cockroach/pkg/cmd/roachtest/tests.registerYCSB.func1
          |     github.com/cockroachdb/cockroach/pkg/cmd/roachtest/tests/ycsb.go:92
          | github.com/cockroachdb/cockroach/pkg/cmd/roachtest/tests.registerYCSB.func4
          |     github.com/cockroachdb/cockroach/pkg/cmd/roachtest/tests/ycsb.go:130
          | main.(*testRunner).runTest.func2
          |     main/pkg/cmd/roachtest/test_runner.go:930
        Wraps: (2) monitor failure
        Wraps: (3) attached stack trace
          -- stack trace:
          | main.(*monitorImpl).wait.func2
          |     main/pkg/cmd/roachtest/monitor.go:171
        Wraps: (4) monitor task failed
        Wraps: (5) attached stack trace
          -- stack trace:
          | main.init
          |     main/pkg/cmd/roachtest/monitor.go:80
          | runtime.doInit
          |     GOROOT/src/runtime/proc.go:6340
          | runtime.main
          |     GOROOT/src/runtime/proc.go:233
          | runtime.goexit
          |     GOROOT/src/runtime/asm_amd64.s:1594
        Wraps: (6) t.Fatal() was called
        Error types: (1) *withstack.withStack (2) *errutil.withPrefix (3) *withstack.withStack (4) *errutil.withPrefix (5) *withstack.withStack (6) *errutil.leafError

Parameters: ROACHTEST_cloud=aws , ROACHTEST_cpu=32 , ROACHTEST_encrypted=false , ROACHTEST_ssd=0

Help

See: [roachtest README](https://github.com/cockroachdb/cockroach/blob/master/pkg/cmd/roachtest/README.md) See: [How To Investigate \(internal\)](https://cockroachlabs.atlassian.net/l/c/SSSBr8c7)

/cc @cockroachdb/test-eng

This test on roachdash | Improve this report!

Jira issue: CRDB-21142

[erikgrinaker]

Failed to execute hostname. Exit code 255 indicates an SSH failure.

@cockroachdb/test-eng Anything to look into here, or should we just close this out as infra flake?

3: pghosts: GetInternalIP: failed to execute hostname on teamcity-7288542-1667452850-08-n4cpu32:4:: exit status 255

[srosenberg]

@erikgrinaker Thanks for triaging! Before we close out, we should take a look at the following,

• why wasn't 255 redirected to the ssh catch-all issue?
• GetInternalIP is using hostname --all-ip-addresses and relying on the order which the manual explicitly says not to.(If the host has multiple network interfaces, it may end up returning the wrong ip.)

@smg260 please have a look

[smg260]

SSH flake error redirection is in master, but even that version only handles user initiated commands, so would miss this. However, the SSH retry PR will catch this (won't retry this), as it has moved the error classification into session.go, to which all commands are routed.

[erikgrinaker]

Opened #98285, closing as infra flake.