15 thoughts on - CentOS 7 (1511) Is Released

• I just updated to 7.2 from 7.1. I did lsb_release -a and it says 7.2.1511. I haven’t rebooted yet, which items would run with new binaries, anything that isn’t running yet? Ssay I had apache running, it wouldn’t pick up new apache until a reboot, right?

  Thanks,

  -wes

• I have no idea, but there are security kernel updates:

  https://lists.CentOS.org/pipermail/CentOS-cr-announce/2015-November/002347.html

  https://lists.CentOS.org/pipermail/CentOS-cr-announce/2015-December/002732.html

  And those will not be active without a reboot.

• most service updates will restart the service

  —
  john r pierce, recycling bits in santa cruz

• Thanks to you and John R Pierce for your replies.

  -wes

• You can always use the “needs-restarting” script to see what you need to restart.

  [Someone suggested “lsof | grep DEL | grep /usr” as an alternative. I haven’t used that approach or compared it to “needs-restarting” so don’t know exactly which is a better approach.]

  With an update from one point release to another I would think that you’d have a rather unstable system until you do a reboot.

• Hi,

  the recommended way is to install yum-plugin-ps and use the ‘yum ps’
  command…

  //Zdenek

• Well the point seems rather moot for the present update as there is no way you should not reboot after the milestone of a point release …

• No discrepancy here ;-)

  I just wanted to share this because some of the plug-ins didn’t exist before and therefore not everyone is aware of them…

  //Zdenek

• Will they? That sound like a pretty terrible idea.

  Regards,
  Dennis

• No they don’t (opposite to e.g. Debian).

  //Zdenek

• Am 15.12.2015 um 14:31 schrieb Zdenek Sedlak :

  If updated via rpm/yum, they do a “condrestart”, but OPs context is an 7.1503 to 7.1511 “upgrade”. Thus, the execution environment changes significantly (glibc, kernel etc.) and this should be addressed with a reboot. If in the future only a “service” (e.g. httpd) gets an update, then rpm handles the restarting process of the running service.

• Thank for the explanation. When did this change? I always believed the running daemon is not touched by rpm/yum…

  IMHO this is a dangerous behaviour because of possible configuration changes which need to be merged first…

  //Zdenek

• The change was made years ago. From the lack of screaming, you can tell that rpm is being reasonably clever about it.

  Apache, for example, checks its config syntax before restarting: man apachectl, see ‘graceful’ and ‘restart’.

  — greg

• I can find no evidence that rpm does anything of the sort. httpd restarts on upgrade because its maintainers included a command to do so in the %posttrans scriptlet.

  httpd checks its configuration syntax before *reloading* which is not what happens during an rpm upgrade. An upgrade will run “systemctl try-restart”, which stops and starts the service if it is running.

• I can think of another example of a package restarting a service:

  The glibc and glibc-common packages run a command called
  /usr/sbin/glibc_post_upgrade.x86_64 or
  /usr/sbin/glibc_post_upgrade.i686 (depending on whether its an x86_64
  or i686 package) in their RPM postinstall scriptlet.

  It runs service sshd condrestart if it detects that
  /etc/rc.d/init.d/sshd exists (so, nothing on systemd-based systems). It also runs ‘telinit u’ too.