hi,
I'd really like to have something where a user, on login to the vm or machine instance is told 'you have XX pending updates'. the data can come from a yum-plugin writing something somewhere ( so its not a case of running yum on each login ).
Who wants to take a stab at writing something for this ?
- KB
On 06/12/2014 12:38 PM, Karanbir Singh wrote:
hi,
I'd really like to have something where a user, on login to the vm or machine instance is told 'you have XX pending updates'. the data can come from a yum-plugin writing something somewhere ( so its not a case of running yum on each login ).
Who wants to take a stab at writing something for this ?
- KB
For CentOS 6 there are gpk-update-viewer and gpk-update-icon commands.
gpk-update-icon is for noticication there is something to update (orange 8-point star), that command can be added to Startup Applications, and gpk-update-viewer is what you get when you click on the star.
have you had something else on the mind? Or for CentOS 7 perhaps?
On 06/12/2014 01:16 PM, Ljubomir Ljubojevic wrote:
On 06/12/2014 12:38 PM, Karanbir Singh wrote:
hi,
I'd really like to have something where a user, on login to the vm or machine instance is told 'you have XX pending updates'. the data can come from a yum-plugin writing something somewhere ( so its not a case of running yum on each login ).
Who wants to take a stab at writing something for this ?
- KB
For CentOS 6 there are gpk-update-viewer and gpk-update-icon commands.
gpk-update-icon is for noticication there is something to update (orange 8-point star), that command can be added to Startup Applications, and gpk-update-viewer is what you get when you click on the star.
have you had something else on the mind? Or for CentOS 7 perhaps?
mostly looking to target VM's and cloudinstances that have barely enough to get yum and openssh-server running, so a yum plugin with an output that sends to motd work work best
On 12/06/14 14:11, Karanbir Singh wrote:
On 06/12/2014 01:16 PM, Ljubomir Ljubojevic wrote:
On 06/12/2014 12:38 PM, Karanbir Singh wrote:
hi,
I'd really like to have something where a user, on login to the vm or machine instance is told 'you have XX pending updates'. the data can come from a yum-plugin writing something somewhere ( so its not a case of running yum on each login ).
Who wants to take a stab at writing something for this ?
- KB
For CentOS 6 there are gpk-update-viewer and gpk-update-icon commands.
gpk-update-icon is for noticication there is something to update (orange 8-point star), that command can be added to Startup Applications, and gpk-update-viewer is what you get when you click on the star.
have you had something else on the mind? Or for CentOS 7 perhaps?
mostly looking to target VM's and cloudinstances that have barely enough to get yum and openssh-server running, so a yum plugin with an output that sends to motd work work best
But that needs thought about how to not overwrite user customised motd files. Ubuntu (yeah, I know) have a /etc/update-motd.d directory which writes the motd from the component scripts in that directory but a 5 minute search didn't make it easay for me to find out how it gets invoked.
T
On 06/12/2014 08:28 AM, Trevor Hemsley wrote:
On 12/06/14 14:11, Karanbir Singh wrote:
On 06/12/2014 01:16 PM, Ljubomir Ljubojevic wrote:
On 06/12/2014 12:38 PM, Karanbir Singh wrote:
hi,
I'd really like to have something where a user, on login to the vm or machine instance is told 'you have XX pending updates'. the data can come from a yum-plugin writing something somewhere ( so its not a case of running yum on each login ).
Who wants to take a stab at writing something for this ?
- KB
For CentOS 6 there are gpk-update-viewer and gpk-update-icon commands.
gpk-update-icon is for noticication there is something to update (orange 8-point star), that command can be added to Startup Applications, and gpk-update-viewer is what you get when you click on the star.
have you had something else on the mind? Or for CentOS 7 perhaps?
mostly looking to target VM's and cloudinstances that have barely enough to get yum and openssh-server running, so a yum plugin with an output that sends to motd work work best
But that needs thought about how to not overwrite user customised motd files. Ubuntu (yeah, I know) have a /etc/update-motd.d directory which writes the motd from the component scripts in that directory but a 5 minute search didn't make it easay for me to find out how it gets invoked.
T _______________________________________________ CentOS-devel mailing list CentOS-devel@centos.org http://lists.centos.org/mailman/listinfo/centos-devel
I used to admin a few of those, so I know where to look:
https://wiki.ubuntu.com/UpdateNotifier http://packages.ubuntu.com/search?keywords=update-notifier
Pat
On 12/06/14 14:30, Pat Riehecky wrote:
On 06/12/2014 08:28 AM, Trevor Hemsley wrote:
On 12/06/14 14:11, Karanbir Singh wrote:
On 06/12/2014 01:16 PM, Ljubomir Ljubojevic wrote:
On 06/12/2014 12:38 PM, Karanbir Singh wrote:
hi,
I'd really like to have something where a user, on login to the vm or machine instance is told 'you have XX pending updates'. the data can come from a yum-plugin writing something somewhere ( so its not a case of running yum on each login ).
Who wants to take a stab at writing something for this ?
- KB
For CentOS 6 there are gpk-update-viewer and gpk-update-icon commands.
gpk-update-icon is for noticication there is something to update (orange 8-point star), that command can be added to Startup Applications, and gpk-update-viewer is what you get when you click on the star.
have you had something else on the mind? Or for CentOS 7 perhaps?
mostly looking to target VM's and cloudinstances that have barely enough to get yum and openssh-server running, so a yum plugin with an output that sends to motd work work best
But that needs thought about how to not overwrite user customised motd files. Ubuntu (yeah, I know) have a /etc/update-motd.d directory which writes the motd from the component scripts in that directory but a 5 minute search didn't make it easay for me to find out how it gets invoked.
T _______________________________________________ CentOS-devel mailing list CentOS-devel@centos.org http://lists.centos.org/mailman/listinfo/centos-devel
I used to admin a few of those, so I know where to look:
https://wiki.ubuntu.com/UpdateNotifier http://packages.ubuntu.com/search?keywords=update-notifier
Pat
I checked further and older ubuntu releases used to use update-motd but they've now modified pam_motd to invoke the scripts in /etc/update-motd.d on login instead. I checked my RHEL7rc VM and those modifications are not present there.
T
On 06/12/2014 02:45 PM, Trevor Hemsley wrote:
I checked further and older ubuntu releases used to use update-motd but they've now modified pam_motd to invoke the scripts in /etc/update-motd.d on login instead. I checked my RHEL7rc VM and those modifications are not present there.
this is just a super complex solution to a problem that is far simpler than it seems.
if a yum-plugin is able to drop the number somewhere, an echo in .skel would be all the 'UI' we need, and if its there for the default user on instance bringup, the skel should get copied over ( and then the user can do whatever they please ) to all new users as well.
would that be all we need here ?
Hello Karanbir,
On Thu, Jun 12, 2014 at 03:01:14PM +0100, Karanbir Singh wrote:
On 06/12/2014 02:45 PM, Trevor Hemsley wrote:
I checked further and older ubuntu releases used to use update-motd but they've now modified pam_motd to invoke the scripts in /etc/update-motd.d on login instead. I checked my RHEL7rc VM and those modifications are not present there.
this is just a super complex solution to a problem that is far simpler than it seems.
if a yum-plugin is able to drop the number somewhere, an echo in .skel would be all the 'UI' we need, and if its there for the default user on instance bringup, the skel should get copied over ( and then the user can do whatever they please ) to all new users as well.
would that be all we need here ?
Maybe limit the time this information is presented: only show the nformation if the number of updates pending was verified within the last 12/24 hours.
best regards,
Florian La Roche
On Thu, Jun 12, 2014 at 9:01 AM, Karanbir Singh mail-lists@karan.org wrote:
I checked further and older ubuntu releases used to use update-motd but they've now modified pam_motd to invoke the scripts in /etc/update-motd.d on login instead. I checked my RHEL7rc VM and those modifications are not present there.
this is just a super complex solution to a problem that is far simpler than it seems.
if a yum-plugin is able to drop the number somewhere, an echo in .skel would be all the 'UI' we need, and if its there for the default user on instance bringup, the skel should get copied over ( and then the user can do whatever they please ) to all new users as well.
would that be all we need here ?
Minor warning: if you are using backuppc to back up a host with rsync over ssh, it will fail if there is any output before the rsync startup. Native rsync will ignore it, but not the backuppc server's rsync-in-perl implementation. So, randomly adding messages at login may have side effects.
On 06/12/2014 09:22 PM, Les Mikesell wrote:
On Thu, Jun 12, 2014 at 9:01 AM, Karanbir Singh mail-lists@karan.org wrote:
I checked further and older ubuntu releases used to use update-motd but they've now modified pam_motd to invoke the scripts in /etc/update-motd.d on login instead. I checked my RHEL7rc VM and those modifications are not present there.
this is just a super complex solution to a problem that is far simpler than it seems.
if a yum-plugin is able to drop the number somewhere, an echo in .skel would be all the 'UI' we need, and if its there for the default user on instance bringup, the skel should get copied over ( and then the user can do whatever they please ) to all new users as well.
would that be all we need here ?
Minor warning: if you are using backuppc to back up a host with rsync over ssh, it will fail if there is any output before the rsync startup. Native rsync will ignore it, but not the backuppc server's rsync-in-perl implementation. So, randomly adding messages at login may have side effects.
does it fail, with enough context to figure out what happened ? if so, we then just need to make it easy to turn off.
On Thu, Jun 12, 2014 at 3:25 PM, Karanbir Singh mail-lists@karan.org wrote:
Minor warning: if you are using backuppc to back up a host with rsync over ssh, it will fail if there is any output before the rsync startup. Native rsync will ignore it, but not the backuppc server's rsync-in-perl implementation. So, randomly adding messages at login may have side effects.
does it fail, with enough context to figure out what happened ? if so, we then just need to make it easy to turn off.
I think it tries to interpret the ascii value of the text as an rsync version number and logs it as a mismatch. So not exactly a 'user friendly' error message, and since backuppc normally runs unattended you may not find out about it until several days have passed - and then only if your email setup works.
But,if it only displays on interactive logins (where ssh doesn't pass a command) it should be OK.
On 06/12/2014 03:25 PM, Karanbir Singh wrote:
On 06/12/2014 09:22 PM, Les Mikesell wrote:
On Thu, Jun 12, 2014 at 9:01 AM, Karanbir Singh mail-lists@karan.org wrote:
I checked further and older ubuntu releases used to use update-motd but they've now modified pam_motd to invoke the scripts in /etc/update-motd.d on login instead. I checked my RHEL7rc VM and those modifications are not present there.
this is just a super complex solution to a problem that is far simpler than it seems.
if a yum-plugin is able to drop the number somewhere, an echo in .skel would be all the 'UI' we need, and if its there for the default user on instance bringup, the skel should get copied over ( and then the user can do whatever they please ) to all new users as well.
would that be all we need here ?
Minor warning: if you are using backuppc to back up a host with rsync over ssh, it will fail if there is any output before the rsync startup. Native rsync will ignore it, but not the backuppc server's rsync-in-perl implementation. So, randomly adding messages at login may have side effects.
does it fail, with enough context to figure out what happened ? if so, we then just need to make it easy to turn off.
Similarly, putting the content in /etc/motd may have issues.
A few remote connection protocols offer up /etc/motd before users authenticate. Sending out to anonymous users "You've got 300 un-applied updates" might be bad.
Pat
On 2014-06-12 22:36, Pat Riehecky wrote:
On 06/12/2014 03:25 PM, Karanbir Singh wrote:
On 06/12/2014 09:22 PM, Les Mikesell wrote:
Minor warning: if you are using backuppc to back up a host with rsync over ssh, it will fail if there is any output before the rsync startup. Native rsync will ignore it, but not the backuppc server's rsync-in-perl implementation. So, randomly adding messages at login may have side effects.
does it fail, with enough context to figure out what happened ? if so, we then just need to make it easy to turn off.
Similarly, putting the content in /etc/motd may have issues.
A few remote connection protocols offer up /etc/motd before users authenticate. Sending out to anonymous users "You've got 300 un-applied updates" might be bad.
Pat
Isn't that what /etc/issue is for? Sounds like something that should be reported as a bug and/or worked around regardless.
/etc/motd is not displayed on non-interactive ssh logins (at least not in the one case I have available to test with).
On 12 June 2014 08:01, Karanbir Singh mail-lists@karan.org wrote:
On 06/12/2014 02:45 PM, Trevor Hemsley wrote:
I checked further and older ubuntu releases used to use update-motd but they've now modified pam_motd to invoke the scripts in /etc/update-motd.d on login instead. I checked my RHEL7rc VM and those modifications are not present there.
this is just a super complex solution to a problem that is far simpler than it seems.
if a yum-plugin is able to drop the number somewhere, an echo in .skel would be all the 'UI' we need, and if its there for the default user on instance bringup, the skel should get copied over ( and then the user can do whatever they please ) to all new users as well.
So a cron job which does a
yum list updates | awk 'BEGIN{x=0}; {if ($NF~'update') {x=x+1}}; END{print "Your system is behind " x " updates."}' > /etc/centos-updates Your system is behind 37 updates.
with it all being in python or something. Then in .bashrc a bit of 'if tty && SEENALREADY then cat /etc/centos-updates; SEENALREADY=1; fi' type logic
On Thu, Jun 12, 2014 at 4:36 PM, Stephen John Smoogen smooge@gmail.com wrote:
if a yum-plugin is able to drop the number somewhere, an echo in .skel would be all the 'UI' we need, and if its there for the default user on instance bringup, the skel should get copied over ( and then the user can do whatever they please ) to all new users as well.
So a cron job which does a
yum list updates | awk 'BEGIN{x=0}; {if ($NF~'update') {x=x+1}}; END{print "Your system is behind " x " updates."}' > /etc/centos-updates Your system is behind 37 updates.
with it all being in python or something. Then in .bashrc a bit of 'if tty && SEENALREADY then cat /etc/centos-updates; SEENALREADY=1; fi' type logic
a) shouldn't this have a concept of security-related vs other updates? b) maybe use .bash_profile vs. .bashrc? c) should it handle multiple users and notify each once? c) isn't there something gnome-ish that already does this with a gui? Can the non-gui version share the check and pending value?
On 12 June 2014 15:46, Les Mikesell lesmikesell@gmail.com wrote:
On Thu, Jun 12, 2014 at 4:36 PM, Stephen John Smoogen smooge@gmail.com wrote:
if a yum-plugin is able to drop the number somewhere, an echo in .skel would be all the 'UI' we need, and if its there for the default user on instance bringup, the skel should get copied over ( and then the user can do whatever they please ) to all new users as well.
So a cron job which does a
yum list updates | awk 'BEGIN{x=0}; {if ($NF~'update') {x=x+1}};
END{print
"Your system is behind " x " updates."}' > /etc/centos-updates Your system is behind 37 updates.
with it all being in python or something. Then in .bashrc a bit of 'if
tty
&& SEENALREADY then cat /etc/centos-updates; SEENALREADY=1; fi' type
logic
a) shouldn't this have a concept of security-related vs other updates? b) maybe use .bash_profile vs. .bashrc? c) should it handle multiple users and notify each once? c) isn't there something gnome-ish that already does this with a gui? Can the non-gui version share the check and pending value?
That is a completely different set of requirements than what was given by KB.
a) Security updates would require that a yum plugin is installed to parse that information. b) Correct. It probably would go in /etc/profile.d or some such thing. c) The yum is to be run by a cron job to meet KB's requirement that yum is not run every time someone logs in. The login profile fires up if the shell is interactive and the environment variable to not show it again is not set. d) The GNOME/KDE gui uses packagekit to basically do a repoquery update every couple of minutes. . it is also a bunch of packages that a remote vm probably won't have installed.
On 06/12/2014 06:11 AM, Karanbir Singh wrote:
On 06/12/2014 01:16 PM, Ljubomir Ljubojevic wrote:
On 06/12/2014 12:38 PM, Karanbir Singh wrote:
hi,
I'd really like to have something where a user, on login to the vm or machine instance is told 'you have XX pending updates'. the data can come from a yum-plugin writing something somewhere ( so its not a case of running yum on each login ).
Who wants to take a stab at writing something for this ?
- KB
For CentOS 6 there are gpk-update-viewer and gpk-update-icon commands.
gpk-update-icon is for noticication there is something to update (orange 8-point star), that command can be added to Startup Applications, and gpk-update-viewer is what you get when you click on the star.
have you had something else on the mind? Or for CentOS 7 perhaps?
mostly looking to target VM's and cloudinstances that have barely enough to get yum and openssh-server running, so a yum plugin with an output that sends to motd work work best
Are you referring to something like "update-motd"? That appears to be installed on my Amazon EC2 instances; when I log in, I get a message like:
1 package(s) needed for security, out of 3 available Run "sudo yum update" to apply all updates.
The RPM "update-motd" (in Amazon's repo) claims the license is ASL 2.0.
I found an old article at: http://www.linuxjournal.com/content/tech-tip-periodically-update-your-motd-u...
-Greg
On 06/12/2014 02:30 PM, Greg Bailey wrote:
Are you referring to something like "update-motd"? That appears to be installed on my Amazon EC2 instances; when I log in, I get a message like:
1 package(s) needed for security, out of 3 available Run "sudo yum update" to apply all updates.
The RPM "update-motd" (in Amazon's repo) claims the license is ASL 2.0.
I found an old article at: http://www.linuxjournal.com/content/tech-tip-periodically-update-your-motd-u...
something like that would / could work - are the sources handy ?
On 12/06/14 15:02, Karanbir Singh wrote:
On 06/12/2014 02:30 PM, Greg Bailey wrote:
Are you referring to something like "update-motd"? That appears to be installed on my Amazon EC2 instances; when I log in, I get a message like:
1 package(s) needed for security, out of 3 available Run "sudo yum update" to apply all updates.
The RPM "update-motd" (in Amazon's repo) claims the license is ASL 2.0.
I found an old article at: http://www.linuxjournal.com/content/tech-tip-periodically-update-your-motd-u...
something like that would / could work - are the sources handy ?
https://launchpad.net/ubuntu/+archive/primary/+files/update-motd_2.0.orig.ta...
Seems to be the last one from before they butchered pam_motd, tarball contains several things, one update-motd written in python and the original update-motd-orig.sh
T
On Thu, Jun 12, 2014 at 02:11:54PM +0100, Karanbir Singh wrote:
mostly looking to target VM's and cloudinstances that have barely enough to get yum and openssh-server running, so a yum plugin with an output that sends to motd work work best
Also see this RFE: https://bugzilla.redhat.com/show_bug.cgi?id=995537
On 06/12/2014 08:11 AM, Karanbir Singh wrote:
mostly looking to target VM's and cloudinstances that have barely enough to get yum and openssh-server running, so a yum plugin with an output that sends to motd work work best
I took a quick swing at working something up for this, though tuned for centos7, as I used systemd for the timing bits.
Source is at https://github.com/jperrin/update-motd
test packaging is (unsigned) at http://people.centos.org/jperrin/update-motd/
criticism will only be accepted if it comes with a patch :-P
Le 12/06/2014 12:38, Karanbir Singh a écrit :
hi,
I'd really like to have something where a user, on login to the vm or machine instance is told 'you have XX pending updates'. the data can come from a yum-plugin writing something somewhere ( so its not a case of running yum on each login ).
Who wants to take a stab at writing something for this ?
- KB
Hi,
There is nagios plugin to check if a server is up-to-date, it can be useful as an inspiration https://trac.id.ethz.ch/projects/nagios_plugins/wiki/check_updates
Regards
Marianne (who is new on this list)
On Thu, Jun 12, 2014 at 11:38:39AM +0100, Karanbir Singh wrote:
I'd really like to have something where a user, on login to the vm or machine instance is told 'you have XX pending updates'. the data can come from a yum-plugin writing something somewhere ( so its not a case of running yum on each login ). Who wants to take a stab at writing something for this ?
I've been wanting to do this in the Fedora cloud image for quite some time, but haven't had a chance to work on it. See https://fedorahosted.org/cloud/ticket/62 -- hopefully we can collaborate here.
-
Elias Persson -
Florian La Roche -
Greg Bailey -
Jim Perrin -
Karanbir Singh -
Les Mikesell -
Ljubomir Ljubojevic -
Marianne Lombard -
Matthew Miller -
Pat Riehecky -
Stephen John Smoogen -
Trevor Hemsley