On Tue, Jan 21, 2014 at 06:41:04AM -0500, Philippe Clérié wrote: > On 01/21/2014 03:38 AM, Andrei POPESCU wrote: > >On Lu, 20 ian 14, 15:28:33, Henning Follmann wrote: > >> > >>Yes, > >>in Wheezy bootlogd is active by default. In previous versions you had to > >>enable it by setting > >>BOOTLOGD_ENABLE=yes > >>in /etc/default/bootlogd > >> > >>The boot messages will be logged to /var/log/boot > > > >True, but since it's Priority: optional it might not be installed at > >all. > > > >Kind regards, > >Andrei > > > > Ignore my previous message. > > When I started the VM this morning, bootlogd did register several > error lines from gunicorn. Seems it cannot change to the source > directory of one of the sites it's serving. That particular > directory is on a NFS share.
See, that would be information which should be included in your first post. Initially I thought about network issues. Here I use Cisco catalyst switches. And the port take up to 10 seconds to come up. This can cause issues with network services during boot. > > I am going to assume that, on cold boots, when gunicorn starts, NFS > mounts are incomplete because it takes the VM time to get the low > level stuff settled. On warm boots, that work is done, so NFS mounts > complete before gunicorn starts. :-) It's a working hypothesis! > > (Of course that does not explain why bootlogd showed nothing yesterday.) > Well, lets assume that was PEBKAC ;) > But thanks for the suggestions! > -H -- Henning Follmann | hfollm...@itcfollmann.com -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20140121135449.gc12...@newton.itcfollmann.com