timeseries-database

Describe the bug

The worker contains the following code:

for (int i = 0; i < n; i++) {
     Unsafe.getUnsafe().loadFence();
      try {
          try {
               useful |= jobs.get(i).run(workerId);
          } catch (Throwable e) {
               onError(i, e);
          }
        } finally {
          Unsafe.getUnsafe().storeFence();
          }
     }

I be

I'm not sure whether this is a gnocchi problem or oslo_middleware.
We are using Gnocchi in an OpenStack setup (openstack-ansible with Pike) with redis as incoming storage. Recently we had the problem, that redis died and gnocchi didn't answer API requests. a request to metric status was answered by a HTTP 500. Our HAProxy in front of the gnocchi API is configured to query /healthcheck. But it d