x/build/cmd/coordinator: limit number of TryBot reattempts

[mdempsky]

I started http://farmer.golang.org/try?commit=ed6176f8 about 4.5 hours ago, and it appears to have attempted to start a Plan 9 GCE image every 6 minutes (i.e., ~45 times now). While its determination is laudable, I think it could stop after maybe 3.

CC @bradfitz @adg

[adg]

agree

On 23 October 2015 at 09:49, Matthew Dempsky notifications@github.com
wrote:

I started http://farmer.golang.org/try?commit=ed6176f8 about 4.5 hours
ago, and it appears to have attempted to start a Plan 9 GCE image every 6
minutes (i.e., ~45 times now). While its determination is laudable, I think
it could stop after maybe 3.

CC @bradfitz https://github.com/bradfitz @adg https://github.com/adg

—
Reply to this email directly or view it on GitHub
#13026.

[bradfitz]

It used to only try 3. I'm not sure why it's looping forever now. I also can't investigate until Monday or Tuesday, and GCE changes broke Plan 9 anyway, so I just disabled Plan 9 builders for now.

[dmitshur]

CC @golang/release.

[dmitshur]

Fixing this for the rest of the builders involves starting to track the number of retries (for a high-level trybot request on a given repo/commit) due to communication errors, and modifying repeatedCommunicationError to return a non-nil error if that number reaches 3. It's currently a TODO comment in that method:

// TODO: give it two tries at least later (store state
// somewhere; global map?). But for now we're going to
// only give it one try.