bigtable.snippets.writes.writes_test: test_writes failed

[flaky-bot]

This test failed!

To configure my behavior, see the Build Cop Bot documentation.

If I'm commenting on this issue too often, add the buildcop: quiet label and
I will stop commenting.

---

buildID: fadb04c
buildURL: Build Status, Sponge
status: failed

[flaky-bot]

Test passed in build 63e99d6 (Build Status, Sponge)! Closing this issue.

[flaky-bot]

Looks like this issue is flaky. 😟

I'm going to leave this open and stop commenting.

A human should fix and close this.

---

commit: ae464b1
buildURL: Build Status, Sponge
status: failed

[tmatsuo]

From the log
row.commit() ended up with google.api_core.exceptions.DeadlineExceeded: 504 Deadline Exceeded

@billyjacobson Does it make sense to update the sample to handle this error and retry few times? If not, maybe we should wrap the each sample invocation with retries.

Because now test_writes has 4 consecutive sample invocation.

When the server is little bit flaky, the success rate of the test is significantly lowered. Let's say, the rpc success rate is 0.9. Then the success rate of test_writes will be 0.9^4 = 0.6561.

What if we retry this test 3 times? The success rate will be:
1 - 0.3439^3 = ~0.959

However, if we wrap these sample invocations with 3 times retry, the success rate of a single sample will be:
1-0.1^3 = 0.999

The success rate of the whole test will be:
0.999^4 = ~0.996

It's significantly higher.