Red Hat Bugzilla – Bug 135114
autofs init script start/restart doesn't check for existing busy mounts
Last modified: 2007-11-30 17:07:04 EST
From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7) Gecko/20040823 Description of problem: When "service autofs stop" or "service autofs restart" is run, busy mounts do not get stopped. When the "start" function is called, it does not check to see if any of these busy mounts still exist. The result is that "service autofs restart" frequently causes two automount processes to operate on the same mount point, which can result in an unstable system. Version-Release number of selected component (if applicable): autofs-4.1.3-12 How reproducible: Always Steps to Reproduce: 1. set up an automount controlled directory auto.master contains: /home /etc/auto.home auto.home contains: zillabug nfsserver:/home/zillabug 2. service autofs start 3. cd /home/zillabug 4. service autofs stop the automount process for /home is still running at this point 5. service autofs start Actual Results: There are now two automount processes trying to control /home and everything underneath it. Expected Results: the initscript (in step 5) should realize that /home is already controlled by automount, and refuse to start that particular map. Additional info:
Created attachment 104956 [details] My suggested fix for this bug
This is something that the daemon should already handle. This code exists in the automount startup: if ((ap.state != ST_INIT) || is_mounted(path)) { /* This can happen if an autofs process is already running*/ error("mount_autofs: already mounted"); return -1; } Are there any autofs messages in /var/log/messages? What do /proc/mounts and /etc/mtab look like both before and after the subsequent autofs start?
My mistake. The server I tested this on must have still been running autofs 3. After reading this, I went to a few servers running autofs 4 and was not able to reproduce.