Try deploying big cluster

100 is enough for now, I guess.

Attach logs if something fails.

changed milestone to %Picodata 2023-06

changed milestone to %Release 0.1 - Demo

assigned to @sv

added backlog label

removed backlog label

added in-progress label

Запускал у себя на ноуте на версии 0c5b947f так:

diff --git a/test/int/conftest.py b/test/int/conftest.py
index 247ad01..796d6b3 100644
--- a/test/int/conftest.py
+++ b/test/int/conftest.py
@@ -299,7 +299,7 @@ class Instance:

         assert want == have

-    @funcy.retry(tries=20, timeout=0.1)
+    @funcy.retry(tries=120, timeout=1)
     def wait_ready(self):
         status = self._raft_status()
         assert status.is_ready
diff --git a/test/int/test_big.py b/test/int/test_big.py
new file mode 100644
index 0000000..9177752
--- /dev/null
+++ b/test/int/test_big.py
@@ -0,0 +1,2 @@
+def test_big(cluster):
+    cluster.deploy(instance_count=99)

Кластер за 2 минуты так и не собрался. Логи: big_test.log.gz При этом CPU usage было < 50%.

На 60 инстансах один раз собрался за секунд 25-30.

closed

removed in-progress label