Control: clone -1 -2 Control: retitle -2 tgt: init script stop action should always stop tgtd Control: reopen -2 Control: found -2 1:1.0.50-1
On 19:17 Wed 17 Sep , Patrick Matthäi wrote: > Am 16.09.2014 um 11:55 schrieb Apollon Oikonomopoulos: > >> a) stop vs. forcedstop > >> stop does not work if there are still open sessions: > >> # /etc/init.d/tgt stop ; echo $? > >> [FAIL] Stopping target framework daemon: tgtd failed! > >> Some initiators are still connected - could not stop tgtd > >> 2 > >> I had to change the use of stop() to forcedstop(), so that tgt works with > >> HeartBeat. > > > > This is intentional and I am a bit reluctant to change the behavior at > > this point (although under systemd we default to forcedstop, having no > > ability to define custom actions). I'll think a bit more about > > defaulting to a hard behavior before the next release. Note that failing > > in this case does not seem to be an LSB violation. > > Think about stopping a SQL or Webservice, where are still (and any time) > users. Not stopping on stop is buggy :( You're right. However HTTP or MySQL clients might be able to deal with this in a graceful manner, whereas a system whose root fs resides on an iSCSI disks will probably not be excited if you pull the root fs under its feet. Note that I don't like stop vs forcedstop either and as I said we already force-stop under systemd by default, so I'll probably change it. For the time being I'm cloning this as a different bug report and leaving it open. Thanks again, Apollon -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
