roachtest: fix a recently-introduced logging bug

[andreimatei]

Release note: None

[cockroach-teamcity]

This change is 

[a-robinson]

Ah, was this the cause of all the nil pointer exception roachtest failures this afternoon?

[andreimatei]

@a-robinson yes, very likely (if the NPE is in logging).
I've seen cause I was playing around myself, but I now see a couple of failures in my inbox too.
I'm confused about why these tests were running (presumably on TC) in the afternoon (and how come they use a roachtest binary from earlier today); in any way sorry for the trouble.

[benesch]

The roach test run takes 16h or so. It's last night's run that you're seeing.

[benesch]

Never mind. It's the "nightly" build that Alex kicked off to test rc2.

[andreimatei]

bors r+

…

On Wed, Oct 24, 2018 at 3:31 PM Nikhil Benesch ***@***.***> wrote: Never mind. It's the "nightly" build that Alex kicked off to test rc2. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#31822 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAXBcQbAUProjgzGrQvlUOfv1zSySZQGks5uoMAbgaJpZM4X4oUP> .

[craig]

Build succeeded

• GitHub CI (Cockroach)