Changes between Version 12 and Version 13 of GEC24Agenda/DeveloperRoundtable

03/11/16 13:21:48 (5 years ago)



  • GEC24Agenda/DeveloperRoundtable

    v12 v13  
    5151  - How can Jacks be extended to handle new types of resources?
    53 If you would like to add a topic, please edit this wiki page or email [ Tom Mitchell].
     53== Summary ==
     55 * GENI Developer collaboration
     57  The sense of the room was that the geni-developers google group was
     58  working for collaboration and discussion. While some issues had been
     59  left unresolved, those who raised the topics felt the discussion was
     60  valuable and had chosen not to pursue resolution.
     62  There was a desire to have a listing of GENI-related code
     63  repositories. It was pointed out that this came up at the GEC 23
     64  Developer's Roundtable and that a [wiki:GENIDeveloper/Repositories wiki page] was created to
     65  list code repositories.
     67 * Extensions to APIs
     69  The !CloudLab team discussed a new capability to dynamically grow and
     70  shrink slices. This is a capability for slices in !CloudLab that can
     71  have nodes added during light usage and nodes taken away during heavy
     72  usage. This capability is different from the slice update API call
     73  and different from the update capability demonstrated by the ExoGENI
     74  team via Flukes.
     76 * Experiment management with Chef
     78  Dmitry Duplyakin, of the !CloudLab team gave an
     79  overview of his use of Chef to configure experiments. Chef provides a
     80  function similar to Ansible which is used in some GENI tutorials. The
     81  sense of the room was that sharing knowledge about multiple tools was
     82  useful and that there is no need to standardize on a single tool
     83  within the GENI community. Dmitry also pointed out the
     84  [ Emulab GitHub area] which includes a
     85  [ Chef repository] containing
     86  cookbooks he has written.
     88 * Experiment management
     90  Rob Ricci and Nick Bastin led a discussion about possible
     91  modifications to request RSpecs to allow passing shared secrets,
     92  like API keys and similar tokens, through the aggregate manager to be
     93  available within a slice via geni-get. The ExoGENI folks discussed
     94  how the NDL translator might handle the additional information. There
     95  was also a discussion about appropriate XML namespaces for this
     96  information. The !CloudLab team will do some prototyping and
     97  coordinate with Nick and the ExoGENI team.
     99 * geni-lib development
     101  The !CloudLab team is actively working with and contributing to
     102  geni-lib. There was a discussion of desirable features and what
     103  should go into a geni-lib 1.0 release. For those wishing to follow
     104  along, geni-lib development happens on [ BitBucket].
     106 * !OpenFlow support
     108  There were no topics discussed related to the
     109  [wiki:GEC24Agenda/OpenFlowSupportExperimenters OpenFlow Support for Experimenters in GENI] session.
     111 * Open issues from  geni-developers[[BR]][[BR]]The Tuesday roundtables wrapped up with discussion about three outstanding threads on the geni-developers google group.[[BR]]
     112    - [!topic/geni-developers/pVntKTpTjq4 Proposal for updated site status/maintenance information via AM API] - Rob Ricci would still like to see this happen in
     113      InstaGENI. He thinks it would be useful to broadcast expected
     114      outages and anticipated heavy load so that tools and users could
     115      see the information.
     116    - [!topic/geni-developers/nOBtrUWGYb0 Add control IP address in the manifest] - Rob Ricci said this
     117      is something they should be able to add to the manifest RSpecs
     118      without too much trouble.
     119    - [!topic/geni-developers/SEVGEWZFbR8 Revealing trust root information in getversion] - There was
     120      agreement that this was useful, but no concrete plans were
     121      developed to put it in place.
     123 * Increasing likelihood of experimenter resource reservations succeeding
     125  Vic Thomas led a discussion about common causes of failure when
     126  trying to reserve resources. Ilya Baldin presented information for
     127  the recent ExoGENI survey of experimenters. Of those who responded
     128  most were able to create their desired topology. Ilya enumerated
     129  various causes of failure that they have seen over the last 18
     130  months as well.
     132  There was discussion about the internal limits on InstaGENI racks
     133  and whether the maximum number of VMs should be adjusted downward
     134  and by how much. Rob Ricci and Hussam Nasir discussed issues with
     135  full disks on the InstaGENI racks, including preventative measures
     136  they could put in place to avoid certain failure modes.
     138 * Monitoring API
     140  There were no additional questions about the monitoring API that was
     141  presented at the
     142  [wiki:GEC24Agenda/MonitoringSupport Monitoring Support for Experimenters and Developers session]
     144 * !CloudLab image service
     146  Rob Ricci gave an overview of a VM image service that the !CloudLab
     147  team is building. Victor Orlikowski is considering a rewrite of the
     148  ExoGENI image service. They discussed common features, how much
     149  overlap there was between the two services, and how they had to
     150  differ to accomodate the differences in the two control frameworks.
     152 * Extending Jacks
     154  There was a brief discussion about extending Jacks to work with new
     155  types of resources. It sounds like some types of resources might be
     156  relatively straightforward to add, but others that use a fairly
     157  different RSpec might be difficult to add inside Jacks proper and
     158  might be better added as a post-processing step.
     160 * RSpec to geni-lib translation
     162  There was a discussion about translating request RSpecs to geni-lib
     163  scripts via code generation. This would allow experimenters to
     164  design their own topology and then see how to build that topology
     165  using geni-lib. They could then use geni-lib to create new
     166  topologies based on the original. The concern was that the generated
     167  code for more complex topologies might not provide a good
     168  programming language to experimenters.