Doxense CI Report for Windows 10

• Commit ID: a84112d
• Result: ✔️ SUCCEEDED
• Build Logs (available for 30 days)

Result of foundationdb-pr-cluster-tests on Linux CentOS 7

• Commit ID: a84112d
• Duration 0:40:14
• Result: ❌ FAILED
• Error: Error while executing command: tar -czf "${CODEBUILD_SRC_DIR}/packaging/docker/website/${FDB_VERSION}/fdb_${FDB_VERSION}.tar.gz" --directory "${CODEBUILD_SRC_DIR}/packaging/docker/website/${FDB_VERSION} .. Reason: exit status 2
• Build Logs (available for 30 days)

Result of foundationdb-pr on Linux CentOS 7

• Commit ID: a84112d
• Duration 0:56:03
• Result: ✅ SUCCEEDED
• Error: N/A
• Build Logs (available for 30 days)

LGTM

Make sure to test this in a HA setup before merging.

I haven't run a Kubernetes test recently (and never explicitly ran a HA test), so it may take a while for me to do this. Is there an alternative way of verifying this? Or, maybe I can file another follow up radar for the HA testing part (if that is the only way of testing this)?

LGTM
Make sure to test this in a HA setup before merging.

I haven't run a Kubernetes test recently (and never explicitly ran a HA test), so it may take a while for me to do this. Is there an alternative way of verifying this? Or, maybe I can file another follow up radar for the HA testing part (if that is the only way of testing this)?

It's probably easier to modify the simulator to use non-numeric DC IDs and reproduce the problem and verify the fix.

Ok, I will try that. Thanks!

It's probably easier to modify the simulator to use non-numeric DC IDs and reproduce the problem and verify the fix.

I applied the following diff:

> @@ -2124,7 +2126,7 @@ void setupSimulatedSystem(std::vector<Future<Void>>* systemActors,
>                                                                           ProcessClass::StatelessClass };
>         for (int dc = 0; dc < dataCenters; dc++) {
>                 // FIXME: test unset dcID
> -               Optional<Standalone<StringRef>> dcUID = StringRef(format("%d", dc));
> +               Optional<Standalone<StringRef>> dcUID = StringRef(format("data%d", dc));
>                 std::vector<UID> machineIdentities;
>                 int machines = machineCount / dataCenters +
>                                (dc < machineCount % dataCenters); // add remainder of machines to first datacenter
> 

The following test causes the sequencer to crash (trying to do "std::stoi()"), on "main", over the above diff:

build_output/bin/fdbserver -r simulation --crash -f /root/src/foundationdb/tests/fast/WriteDuringReadClean.toml -b off -s 22570166

The test succeeds over the changes in this PR. The primary locality id on the sequencer is getting set to "-1" (in this test).

The test succeeds over the changes in this PR. The primary locality id on the sequencer is getting set to "-1" (in this test).

Can you create a PR with your test changes? I think it's worth making the change to catch any future regression.

The test succeeds over the changes in this PR. The primary locality id on the sequencer is getting set to "-1" (in this test).

Can you create a PR with your test changes? I think it's worth making the change to catch any future regression.

I am not sure we can check in this change. Check the top comment in "SimulationConfig::setRegions()" - maybe there are similar assumptions at other places too.