From 0166985f96ecfd3ec8ee529c1882cd9abfa1c895 Mon Sep 17 00:00:00 2001
From: Kefu Chai <kchai@redhat.com>
Date: Sat, 2 May 2020 23:54:44 +0800
Subject: [PATCH] qa/tasks/ceph.py: do not use option mimic does not understand

mimic does not have `mon-client-directed-command-retry` option, so we
should not pass this option to a mimic ceph client.

in this change, we fall back to plain retry, if command fails. this
change is not cherry-picked from master. as we don't run upgrade test
from mimic to master.

Signed-off-by: Kefu Chai <kchai@redhat.com>
---
 qa/tasks/ceph.py | 34 ++++++++++++++++++++--------------
 1 file changed, 20 insertions(+), 14 deletions(-)

diff --git a/qa/tasks/ceph.py b/qa/tasks/ceph.py
index aa4c9620fdc52..219c22c6c7458 100644
--- a/qa/tasks/ceph.py
+++ b/qa/tasks/ceph.py
@@ -1680,6 +1680,25 @@ def validate_config(ctx, config):
             last_role = role
 
 
+def stop_logging_health(remote, cluster, retry):
+    # try this several times, since tell to mons is lossy.
+    args = 'sudo ceph --cluster {cluster} {retry_opts} tell mon.* injectargs -- --no-mon-health-to-clog'
+    try:
+        retry_opts = '--mon-client-directed-command-retry {}'.format(retry)
+        remote.run(
+            args=args.format(cluster=cluster,
+                             retry_opts=retry_opts))
+    except run.CommandFailedError:
+        for i in range(retry):
+            try:
+                remote.run(
+                    args=args.format(cluster=cluster,
+                                     retry_opts=''))
+                return
+            except run.CommandFailedError:
+                pass
+
+
 @contextlib.contextmanager
 def task(ctx, config):
     """
@@ -1886,17 +1905,4 @@ def task(ctx, config):
             # a bunch of scary messages unrelated to our actual run.
             firstmon = teuthology.get_first_mon(ctx, config, config['cluster'])
             (mon0_remote,) = ctx.cluster.only(firstmon).remotes.keys()
-            # try this several times, since tell to mons is lossy.
-            mon0_remote.run(
-                args=[
-                    'sudo',
-                    'ceph',
-                    '--cluster', config['cluster'],
-                    '--mon-client-directed-command-retry', '5',
-                    'tell',
-                    'mon.*',
-                    'injectargs',
-                    '--',
-                    '--no-mon-health-to-clog',
-                ]
-            )
+            stop_logging_health(mon0_remote, config['cluster'], 5)
-- 
2.39.5