Innerloop Windows builds always taking two attempts to succeed?

[dagood]

I happened to notice that in the last two internal innerloop builds, the Windows jobs all took exactly two attempts to complete:

https://dev.azure.com/dnceng/internal/_build/results?buildId=2542096&view=results
https://dev.azure.com/dnceng/internal/_build/results?buildId=2542151&view=results

In the build before that, all Windows jobs passed on the first try:

https://dev.azure.com/dnceng/internal/_build/results?buildId=2535458&view=results

I haven't investigated why or what the errors actually were, but wanted to note this down while I'm seeing it. The pattern just seems very odd.

I triggered a few new builds to see if it repros again:

https://dev.azure.com/dnceng/internal/_build/results?buildId=2544949&view=results
https://dev.azure.com/dnceng/internal/_build/results?buildId=2544958&view=results

[dagood]

I think this is just an increase in random flakiness. Both of the newly run jobs failed due to exceeding timeout, with a varying number of retry attempts.