Add garbage collector command to launcher (#20)

* Add garbage collector command to launcher

The seafile garbage collector can be started via the 'gc' command option of the launcher.
This will stop the seafile-server inside the container, then run the 'seaf-gc.sh' script and redirect its output to '/var/log/gc.log'.
Afterwards the whole container will be restarted.

* Move garbage collector execution logic to scripts/gc.sh

The file gc.sh stops the seafile server, starts the garbage collector and restarts the seafile server when the cleanup is finished.
The output of the 'seaf-gc.sh' script is piped to the file '/var/log/gc.log' in append mode.

* Add 'scripts/gc.sh' to watch_controller function in 'scripts/start.py'

The 'start.py' script monitors the seafile-server in a 'watch_controller' function and terminates the container when the server crashed.
However, during a garbage collector cleanup the server needs to be shut down and therefore it is necessary that the 'watch_controller' function only terminates the container if the server is offline and no cleanup is in progress.

* Add the gc command info to README

* Preseve the exit code of seaf-gc.sh

* Fix the perm of scripts/gc.sh

README.md

@@ -92,6 +92,7 @@ Commands:
     enter:      Use docker exec to enter a container
     logs:       Docker logs for container
     rebuild:    Rebuild a container (destroy old, bootstrap, start new)
+    gc:         Start the seafile garbage collector (stops seafile, starts gc, restarts seafile)
 ```
 
 If the environment variable "SUPERVISED" is set to true, the container won't be detached, allowing a process monitoring tool to manage the restart behaviour of the container.

launcher

@@ -11,6 +11,7 @@ usage () {
   echo "    logs:       View the Docker container logs"
   echo "    bootstrap:  Bootstrap the container based on a template"
   echo "    rebuild:    Rebuild the container (destroy old, bootstrap, start new)"
+  echo "    gc:         Start the seafile garbage collector (stops seafile, starts gc, restarts seafile)"
   echo
   echo "Options:"
   echo "    --skip-prereqs             Don't check launcher prerequisites"
@@ -224,6 +225,21 @@ restart() {
     start
 }
 
+gc() {
+    ensure_container_running
+    (
+        loginfo "Stop seafile and start the garbage collector"
+        set +e +x
+        docker exec -it seafile /scripts/gc.sh
+        gc_error=$?
+        if [[ $gc_error -eq 0 ]]; then
+            loginfo "Garbage collector finished!"
+        else
+            err_and_quit "Garbage collector exited with code $gc_error"
+        fi
+    )
+}
+
 check_prereqs() {
     if [[ $SKIP_PREREQS == "true" ]]; then
         return 0
@@ -391,7 +407,7 @@ main() {
     while [[ $# -gt 0 ]]
     do
         case "$1" in
-            bootstrap|start|stop|restart|enter|destroy|logs|rebuild|manual-upgrade)
+            bootstrap|start|stop|restart|enter|destroy|logs|rebuild|manual-upgrade|gc)
                                                 action=${1//-/_}            ; shift 1 ;;
             -h|--help)                          ( usage ; exit 1)           ; shift 1 ;;
             -v|--verbose)                       verbose=true                ; shift 1 ;;

scripts/gc.sh

@@ -0,0 +1,30 @@
+#!/bin/bash
+
+set -e
+
+# Before
+SEAFILE_DIR=/opt/seafile/seafile-server-latest
+
+$SEAFILE_DIR/seafile.sh stop
+
+echo "Waiting for the server to shut down properly..."
+sleep 5
+
+# Do it
+(
+    set +e
+    $SEAFILE_DIR/seaf-gc.sh | tee -a /var/log/gc.log
+    # We want to presevent the exit code of seaf-gc.sh
+    exit "${PIPESTATUS[0]}"
+)
+
+gc_exit_code=$?
+
+# After
+
+echo "Giving the server some time..."
+sleep 3
+
+$SEAFILE_DIR/seafile.sh start
+
+exit $gc_exit_code

scripts/start.py

@@ -25,8 +25,9 @@ def watch_controller():
     maxretry = 4
     retry = 0
     while retry < maxretry:
-        controller_pid = get_command_output('ps aux | grep seafile-controller |grep -v grep || true').strip()
+        controller_pid = get_command_output('ps aux | grep seafile-controller | grep -v grep || true').strip()
-        if not controller_pid:
+        garbage_collector_pid = get_command_output('ps aux | grep /scripts/gc.sh | grep -v grep || true').strip()
+        if not controller_pid and not garbage_collector_pid:
             retry += 1
         else:
             retry = 0