[Planetlab-users] dead nodes

[Jin Liang]

On Mon, 16 May 2005, Steve Muir wrote:

but i guess we can add 
> slice users to the next round of 'node down' emails and see what happens.

This is what I meant. There's no need to send initial emails to
users. But they should receive the last notification that their
slices could be disabled, if still nothing happens. This last
notification is relevant, because it involves their own interest.

Jin


>
>
>
> On Mon, 16 May 2005, Jin Liang wrote:
>
>> Is there any technical difficulty for not notifying
>> the slice owners? As I said the slice owners have
>> the most incentive to cooperate with PlanetLab to
>> solve the problems, I don't see why this cannot be done.
>> 
>> As for UIUC, I THINK (80% sure) I'm also on the tech
>> contact list. Therefore I'm also speaking in general,
>> not about our own site.:)
>> 
>> Thanks!
>> 
>> Jin
>> 
>> On Mon, 16 May 2005, Steve Muir wrote:
>> 
>>> PIs will certainly be notified once nodes pass a certain downtime 
>>> threshold, and it doesn't seem unreasonable to delegate part of the 
>>> responsibility for keeping nodes online to site PIs.  it's also the PIs 
>>> responsbility to make sure that PI and tech contact information is 
>>> up-to-date, even if the PI leaves the institution.  if you think it's 
>>> unfair that you as a slice user get punished for your site's node being 
>>> crappy (i'm speaking generally here, not specifically about you and UIUC) 
>>> make sure your PI knows that and gets on your tech contact's case, or 
>>> volunteer to be tech contact yourself.
>>> 
>>> 
>>> 
>>> On Mon, 16 May 2005, Jin Liang wrote:
>>> 
>>>> Could you please notify the owner of a slice before
>>>> you automatically disable the slice? Slice users
>>>> have more incentive to keep their nodes up and running,
>>>> however, they are often not directly responsible for
>>>> the support issues. Punishing the slice users (silently)
>>>> for the non-responsiveness of the technical contact
>>>> seems unfair, and asking the slice users to periodically
>>>> check their node status may not be feasible (people
>>>> will forget). A notification, however, will allow the
>>>> slice users to help PlanetLab solve the issues.
>>>> 
>>>> Thanks
>>>> 
>>>> Jin
>>>> 
>>>> 
>>>> On Mon, 16 May 2005, Reid Moran wrote:
>>>> 
>>>>> Danny,
>>>>> 
>>>>> I refer you to http://monitor.planet-lab.org/  this site shows the 
>>>>> current status for all Planet Lab nodes.  You will see when you click on 
>>>>> the subsequent pages there is a column within the table titled "RT". This 
>>>>> column shows that there are current support issues for that node, and as 
>>>>> long as there is responsiveness coming from the site about this node, the 
>>>>> site will not be punished.  The goal here is to motivate those sites who 
>>>>> are not responsive in getting their nodes back online. The most likely 
>>>>> scenario will be creating a new status for nodes that are down while the 
>>>>> site responsible is not responding to multiple requests from us to get 
>>>>> those nodes back online.  This is what we are looking to prevent, we do 
>>>>> understand many various circumstances cause nodes to be offline, and will 
>>>>> work with the site to get them back online as long as they are 
>>>>> responsive.
>>>>> 
>>>>> Hope that helps explain.
>>>>> 
>>>>> -Reid
>>>>> 
>>>>> _______________________________________________
>>>>> Users mailing list: Users at lists.planet-lab.org
>>>>> https://lists.planet-lab.org/mailman/listinfo/users
>>>>> 
>>>> 
>>>> _______________________________________________
>>>> Users mailing list: Users at lists.planet-lab.org
>>>> https://lists.planet-lab.org/mailman/listinfo/users
>>>> 
>>> 
>>> _______________________________________________
>>> Users mailing list: Users at lists.planet-lab.org
>>> https://lists.planet-lab.org/mailman/listinfo/users
>>> 
>> 
>