Bart Smaalders wrote:
Todd Pisek wrote:
Nicolas Williams wrote:
On Tue, Jun 09, 2009 at 01:06:01PM -0500, Todd Pisek wrote:
The problem I see with suspend/disable is that if the stop method
fails (there are mounted file systems), then the service remains
disabled and
Your service sho
On Tue, Jun 09, 2009 at 01:09:42PM -0700, johan...@sun.com wrote:
> I don't understand why you've taken this approach. The point is to have
> a sam-qfs service, not an ips-remove method stashed in SMF. When you
> disable the sam-qfs service, why doesn't it unmount the filesystems, if
> that's par
Todd Pisek wrote:
Nicolas Williams wrote:
On Tue, Jun 09, 2009 at 01:06:01PM -0500, Todd Pisek wrote:
The problem I see with suspend/disable is that if the stop method
fails (there are mounted file systems), then the service remains
disabled and
Your service should go into maintenance
Nicolas Williams wrote:
On Tue, Jun 09, 2009 at 11:21:30AM -0700, Bart Smaalders wrote:
Note that system updates (image-update) is always done on an alternate
boot environment (NEVER live), so all of this issues don't occur in
the typical case.
Is there a way to tag a pkg as disallowing update
On Tue, Jun 09, 2009 at 03:03:31PM -0500, Todd Pisek wrote:
> Nicolas Williams wrote:
>> On Tue, Jun 09, 2009 at 01:06:01PM -0500, Todd Pisek wrote:
>>
>>> The problem I see with suspend/disable is that if the stop method
>>> fails (there are mounted file systems), then the service remains
>>>
Nicolas Williams wrote:
On Tue, Jun 09, 2009 at 01:06:01PM -0500, Todd Pisek wrote:
The problem I see with suspend/disable is that if the stop method fails
(there are mounted file systems), then the service remains disabled and
Your service should go into maintenance when disabled and
On Tue, Jun 09, 2009 at 01:06:01PM -0500, Todd Pisek wrote:
> The problem I see with suspend/disable is that if the stop method fails
> (there are mounted file systems), then the service remains disabled and
Your service should go into maintenance when disabled and unable to
quiesce itself. The
On Tue, Jun 09, 2009 at 11:21:30AM -0700, Bart Smaalders wrote:
> Note that system updates (image-update) is always done on an alternate
> boot environment (NEVER live), so all of this issues don't occur in
> the typical case.
Is there a way to tag a pkg as disallowing updates on running images?
_
On Tue, Jun 09, 2009 at 11:21:30AM -0700, Bart Smaalders wrote:
> Note that system updates (image-update) is always done on an alternate
> boot environment (NEVER live), so all of this issues don't occur in the
> typical case.
Though unless he tags his files as needing reboot, then once we start
Todd Pisek wrote:
Yes, to support update, suspend is needed, but so is disable for
uninstall. Even though uninstall is not a common operation in the IPS
model, it needs to be handled.
Right now, rather than kludging up something w/ multiple services, etc,
I recommend that you suggest that p
Danek Duvall wrote:
On Tue, Jun 09, 2009 at 11:12:00AM -0500, Todd Pisek wrote:
:
:
Note that I must not allow IPS to change any package files if there are
mounted QFS file systems (unmount -f is not an option). If there is a way
to selectively replace files via the update mechanism, I be
Danek Duvall wrote:
Bart should confirm, but it looks to me like it'll abort the entire update.
Danek
This is correct.
- Bart
--
Bart Smaalders Solaris Kernel Performance
ba...@cyber.eng.sun.com http://blogs.sun.com/barts
"You will contribute more with mercurial th
On Tue, Jun 09, 2009 at 11:12:00AM -0500, Todd Pisek wrote:
> The old way of thinking with SVR4, our major releases require pkgrm/pkgadd
> (we also have patch sets for older releases).
> What is the IPS replacement for this?
Upgrade is always done by simply installing the new version of the pack
Danek Duvall wrote:
On Tue, Jun 09, 2009 at 08:48:24AM -0500, Todd Pisek wrote:
I'm assuming that for the time being, I can't associate actuators to
services for files that are in the same package as the service manifest and
method. When I try to combine the assembly manifest(s) and method(
On Tue, Jun 09, 2009 at 08:48:24AM -0500, Todd Pisek wrote:
> I'm assuming that for the time being, I can't associate actuators to
> services for files that are in the same package as the service manifest and
> method. When I try to combine the assembly manifest(s) and method(s) with
> the "reg
I'm assuming that for the time being, I can't associate actuators to
services for files that are in the same package as the service manifest
and method. When I try to combine the assembly manifest(s) and method(s)
with the "regular" package, there is a high probability that the
assembly service
16 matches
Mail list logo
