incubator-tashi-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "David O'Hallaron" <d...@cs.cmu.edu>
Subject Re: Destination host unreachable
Date Mon, 01 Aug 2011 17:49:43 GMT
> If the VM is actually up, is the DHCP server capable of handing out an
> address in time for the machine to accept it?

Tashi seems to think the instance is up but it never seems to have
gotten an IP address. After this kind of thing happens, I can
subsequently create a new VM that gets a good IP address, so the DHCP
server is definitely working.

Are you suggesting that the DHCP server might get overloaded, not
respond, and that Tashi doesn't retry the request?

Thanks!
Dave



On Mon, Aug 1, 2011 at 1:35 PM, Michael Stroucken <mxs@cmu.edu> wrote:
> David O'Hallaron wrote:
>>
>> We're running a tashi cluster as the backend for an autograding
>> service at CMU called Autolab. Each time a student hands in their
>> work, the service creates a fresh VM instance, autogrades the students
>> work in this instance, and then destroys it. Now and then, typically
>> in periods of higher load when we have to create multiple instances
>> at the same time, we'll get an instance that can't be reached (see
>> below).
>>
>> Any ideas what might be causing this?
>>
>> Thanks!
>> Dave
>>
>> linux> tashi-client.py getinstances
>>  id  hostId name        user state   disk     memory cores
>> ----------------------------------------------------------
>>  178 6      tango6-rhel gdm  Running rhel.img 512    1
>>
>> linux> ping tango6-rhel.vmnet
>> PING tango6-rhel.vmnet (192.168.1.216) 56(84) bytes of data.
>> From 192.168.1.1 icmp_seq=1 Destination Host Unreachable
>> From 192.168.1.1 icmp_seq=2 Destination Host Unreachable
>>
>
> If the VM is actually up, is the DHCP server capable of handing out an
> address in time for the machine to accept it?
>
> Greetings,
> Michael.
>

Mime
View raw message