[WIP] Jepsen fix (#138)

* Do like redis

* Add workflow to test jepsen tests...

* Remove jepson tests on pull/push

* Let's check a scenario with time-limit 7200

* Change gen/time-limit

* Okay... Let's set it to 3600s

* Take up timeouts!..

* Let's check it with 60000 read timeout

* Rename errors

* Add keys to the mysql hosts

* Fix ssh connection

* fix timeout

* Remove on_push and on_pull_request triggers

Makefile

@@ -46,6 +46,7 @@ jepsen_test:
 	timeout 600 docker exec mysync_mysql2_1 sh -c "/var/lib/dist/base/generate_certs.sh mysync_mysql2_1.mysync_mysql_net && supervisorctl restart mysync && supervisorctl start mysqld"
 	timeout 600 docker exec mysync_mysql3_1 sh -c "/var/lib/dist/base/generate_certs.sh mysync_mysql3_1.mysync_mysql_net && supervisorctl restart mysync && supervisorctl start mysqld"
 	timeout 600 docker exec mysync_mysql1_1 setup.sh
+	timeout 600 bash ./tests/images/copy_keys.sh
 	mkdir -p ./tests/logs
 	(docker exec mysync_jepsen_1 /root/jepsen/run.sh > ./tests/logs/jepsen.log 2>&1 && tail -n 4 ./tests/logs/jepsen.log) || ./tests/images/jepsen_main/save_logs.sh
 	docker compose -p mysync -f ./tests/images/jepsen-compose.yml down --rmi all

tests/images/copy_keys.sh

@@ -0,0 +1,13 @@
+for i in mysql1 mysql2 mysql3; do
+  docker exec mysync_${i}_1 rm /root/.ssh/id_rsa
+  docker exec mysync_${i}_1 ssh-keygen -m PEM -t rsa -b 2048 -N '' -C root@mysync_${i}_1 -f /root/.ssh/id_rsa
+done
+
+for i in mysql1 mysql2 mysql3; do
+  docker cp mysync_${i}_1:/root/.ssh/id_rsa.pub id_rsa.pub
+  for j in mysql1 mysql2 mysql3; do
+    docker exec mysync_${j}_1 /bin/bash -c "echo '$(cat id_rsa.pub)' >>/root/.ssh/authorized_keys"
+  done
+done
+
+rm id_rsa.pub

tests/images/jepsen_common/sshd_config

@@ -5,6 +5,6 @@ X11Forwarding yes
 PrintMotd no
 AcceptEnv LANG LC_*
 Subsystem       sftp    /usr/lib/openssh/sftp-server
-PubkeyAcceptedAlgorithms ssh-rsa
+PubkeyAcceptedAlgorithms +ssh-rsa
 PermitRootLogin yes
 MaxAuthTries 1000000

tests/images/jepsen_main/jepsen/src/jepsen/mysync.clj

@@ -93,15 +93,16 @@
       (try
           (with-conn [c conn]
             (case (:f op)
-              :read (cond (= (count (j/query c ["show slave status for channel ''"])) 0)
+              :read (timeout 60000 (assoc op :type :info, :error "read-timeout")
+                      (cond (= (count (j/query c ["show slave status for channel ''"])) 0)
                           (assoc op :type :ok,
                                     :value (->> (j/query c ["select value from test1.test_set"]
                                                           {:row-fn :value})
                                                 (vec)
                                                 (set)))
                           true
-                          (assoc op :type :info, :error "read-only"))
-              :add (timeout 5000 (assoc op :type :info, :error "timeout")
+                          (assoc op :type :info, :error "read-only")))
+              :add (timeout 5000 (assoc op :type :info, :error "add-timeout")
                     (do
                       (info (str "Adding: " (get op :value) " to " (get c :subname)))
                       (j/execute! c [(str "insert into test1.test_set values ('" (get op :value) "')")])
@@ -266,9 +267,9 @@
                        (fn [] (map gen/once
                                    [{:type :info, :f (rand-nth nemesis-starts)}
                                     {:type :info, :f (rand-nth nemesis-starts)}
-                                    {:type :sleep, :value 120}
+                                    {:type :sleep, :value 60}
                                     {:type :info, :f :stop}
-                                    {:type :sleep, :value 120}])))
+                                    {:type :sleep, :value 60}])))
                      (gen/time-limit 7200))
                 (->> r
                      (gen/stagger 1)