Sliver expiration

[Raj Thakkar]

Hello,

                   I am using stitcher in my experiment. The commands i am using to create sliver is stitcher.py createsliver <slice-name> <rspec name> -o and the sliver get reserved. But i want the sliver for more than 1 day, so i renew the sliver using the command stitcher.py renewsliver <slice-name> <expiration date>. But i keep on losing the sliver after 1 day. So am i missing something? Does renewing sliver has a different command?

Thank you

Regards

Raj Thakkar

[Aaron Helsinger]

renewsliver for resources reserved through stitcher should work just like normal reservations.

Have you looked at the logs from calling 'stitcher.py renewsliver' for errors?

If you cannot tell from the logs what is happening, send us the logs and others can help you read the results.

Aaron

--
GENI Users is a community supported mailing list, so please help by responding to questions you know the answer to.
 
If this is your first time posting a question to this list, please review http://groups.geni.net/geni/wiki/GENIExperimenter/CommunityMailingList
---
You received this message because you are subscribed to the Google Groups "GENI Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to geni-users+...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

[fkandah]

Raj,

Are you loosing the PCs or the VLAN is down? Or are you loosing everything?

[Raj Thakkar]

Hello Aaron

                               The attached files with this e-mail are the errors i am getting when i try and run the renewsliver commands. I am using vlan in my experiment and after 24 hours i lose everything reserved that is PCs and vlan.

Thank you

Regards

Raj Thakkar

--

[Aaron Helsinger]

A few things here:

stitcher.py only knows where to renew your reservations if you run it in the same place where you did your original reservation, and if you haven't cleaned out files that stitcher wrote the last time you ran. That is, when you run stitcher to reserve resources, it creates a file listing the aggregates at which you made reservations.

In your case, that file should be something like "ch.geni.net.MultipathRouting.xyztest-amlist.txt"

If stitcher does not find that file, then it ends up trying to renew your resources at _all_ aggregates. Most of the time, that fails with an error something like "No such slice here", because indeed you have no reservation at that aggregate. Your logs show a lot of those errors, as expected.

Next time, make sure you have that file still, and stitcher will only try to renew your reservation where you have a reservation.

However, event without that file, it should succeed sometimes. And your logs show that it did.
Your logs show that your renewal request succeeded at instageni.rnet.missouri.edu until 2014-02-01
Also at uky.emulab.net

At ION, you got a different error:
"The new expiration time must be at least 15 minutes from the current expiration time!"

The last time of your logs says:
"Renewed slivers on 2 out of 40 aggregates..."

I don't know where you had reservations, so other than that error at ION, I don't know if this is what you expected.

You can always renew explicitly at the aggregates where you have reservations, by specifying the aggregate on the commandline, eg the following command renews resources in your slice until February 4 at 3 different aggregates, showing 3 different ways that can work to identify the aggregate where you want to renew your reservation:

stitcher.py renewsliver xyztest 2014-02-04 -a ion -a urn:publicid:IDN+uky.emulab.net+authority+cm -a https://instageni.rnet.missouri.edu:12369/protogeni/xmlrpc/am/2.0

I hope this helps you,
Aaron

[Aaron Helsinger]

Here's a command that will show you your current sliver expirations at your instageni racks:

stitcher.py -a pg-ky -a https://instageni.rnet.missouri.edu:12369/protogeni/xmlrpc/am/2.0 -V3 status xyztest

You should renew these 2 today to avoid losing them, since your renewsliver you sent renewed them only until 2/1.

And at ION, do this:

stitcher.py -a ion -V2 sliverstatus xyztest

Aaron

[Raj Thakkar]

Hello Aaron,

                                   I tried one of the commands you mentioned regarding renewing sliver. I got an output that the sliver is renewed. The command i used and the output i got is in the attached screenshot.

Thank you

Regards

Raj Thakkar

[Aaron Helsinger]

Looks good!

Each command ended by saying that it had indeed renewed your sliver until the requested time.

Let me know if I can help with anything else.

Aaron

[Raj Thakkar]

Sure Aaron

Thank you

Regards

Raj Thakkar

[Raj Thakkar]

Hello Aaron,

                                     It seems that the command worked fine, but i got an e-mail today, that the sliver is expiring in 8 hours. So i think it didn't got reserved for the until the requested time. The attached screenshot with this e-mail is for the sliver expiration time. Do we need to mention anything else in the command?

Thank you

Regards

Raj Thakkar

[Leigh Stoller]

> It seems that the command worked fine, but i got an e-mail today, that
> the sliver is expiring in 8 hours. So i think it didn't got reserved for
> the until the requested time. The attached screenshot with this e-mail is
> for the sliver expiration time. Do we need to mention anything else in
> the command?

Perhaps a different issue. If you are using Utah (Utah IG Rack, Utah DDC
Rack, or Utah Emulab PG), you are not allowed to let your resources go
idle. There is too much contention for Utah resources, so if you allocate
resources and then do not use them for too long (currently about six
hours), you will lose them cause we will not allow you to renew them.

Lbs

[Raj Thakkar]

I am using Missouri-ig and pg-uky in my experiment.

Thank you

Regards

Raj Thakkar

[Raj Thakkar]

When i used the command of renew sliver it shows that the sliver got renewed until the requested time, but it is actually not renewing the sliver.

Thank you

Regards

Raj Thakkar

[Aaron Helsinger]

Raj-

Is your slice still xyztest in the project MultipathRouting?

You renewed the slivers until midnight UTC on the 5th (you said "2014-02-05"). So about 6 hours from now. That's consistent with the screenshot you showed us.

Leigh:

It looks to me as though the slivers at those aggregates disappeared. The aggregates say they are gone.

Any idea what's happening there?

Aaron

[Leigh Stoller]

> You renewed the slivers until midnight UTC on the 5th (you said "2014-02-05"). So about 6 hours from now. That's consistent with the screenshot you showed us.
>
> Leigh:
>
> It looks to me as though the slivers at those aggregates disappeared. The aggregates say they are gone.
>
> Any idea what's happening there?

See attached screen shot from Missouri. The last renew was done on
the 30th, and it expired 24 hours later. There have been a number
of Create/Delete cycles since then, but no Renews.

This page needs to publicly accessible, since all the log files it
references are also public. Would make it easier then screen shots,
I will put that on my TODO list.

Lbs

[Leigh Stoller]

> See attached screen shot from Missouri.

[Raj Thakkar]

The slice name is abc and it has the same aggregates missouri-ig and pg-uky

Thank you

Regards

Raj Thakkar

On Tue, Feb 4, 2014 at 1:33 PM, Leigh Stoller <lbst...@gmail.com> wrote:

> See attached screen shot from Missouri.

--

[Leigh Stoller]

> The slice name is abc and it has the same aggregates missouri-ig and pg-uky

In that case ... The abc slice still exists at missouri-ig and is set
to expire as Aaron stated. Ditto on pg-uky ...

Lbs