[Seaside] Deploy via...image or monticello?
asqueaker at gmail.com
Tue Jul 6 18:57:53 UTC 2010
If you have HA or can afford a potential service interruption, I think
it is very elegant. Being able to simply push a button and interact
with a full-blown GUI to have assurance of exactly what code is
loaded, along with full interrogation of the running server, that's
pretty nice compared to typing cryptic commands and not necessarily
have that full view.
Having said that, we don't do it at our company (unless an emergency),
because we cannot afford service interruptions. Loading updated
packages could put the running system temporarily in an inconsistent
state of packages whose combination was never tested. Perhaps atomic
loading can help with this, but then you might have a "pause" in
service while multiple packages are loaded and classes recompiled. It
also doesn't address object migrations related to new class-shapes,
Which is why high-availability is a fine tool to have for updates.
Servers of a HA system can be brought down invidually, the software
updated at leisure, and then reinserted into the running node, one at
No service interruption and no panic if something goes wrong while
updating one server; the others in the node are still running. That's
nice too. :)
On Tue, Jul 6, 2010 at 4:32 AM, stephane ducasse
<stephane.ducasse at free.fr> wrote:
> why would not it be elegant?
> On Jul 1, 2010, at 5:14 AM, AxiNat wrote:
>> I always have a VNC server running in the image with the Monticello browser open, so I develop in local, and when I save a new version I just have to check the online image through a VNC client, click at refresh and load the new version. It may not be very elegant but it's very convenient to me, no need to upload files or anything.
>> Bernat Romagosa.
>> seaside mailing list
>> seaside at lists.squeakfoundation.org
> seaside mailing list
> seaside at lists.squeakfoundation.org
More information about the seaside