There are a number of possibilities:
 - There might be some obscure quota issue - all bots use the same quota pool (though quota limits should normally terminate connection very fast)
 - There might be network problems, especially for physical machines (GCE bots shouldn't have this problem - may need cross-checking if we see timeouts on GCE)
 - Something else I can't think of right now?

Local timeouts might not be observed as much because of different network / quota pool / small scale. The scale in particular is important: if timeouts happen once in 10K times, we'll see plenty of them on bots, but hardly any locally. Though I recently saw maybe 1/5 timeout rate on gnumbd - issue 663052.

Adding Infra>Platform component for more input.

This issue has been Available for over a year. If it's no longer important or seems unlikely to be fixed, please consider closing it out. If it is important, please re-triage the issue.

Sorry for the inconvenience if the bug really should have been left as Available. If you change it back, also remove the "Hotlist-Recharge-Cold" label.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot

We have monitoring in place now to tell us when this is particularly bad. I'm going to close this since besides some recent windows slowness that we've fixed and some current limitations with the git protocol that the git server team is addressing things have been steady.