GitLab

seconds), Compare against new one, and skip updating the node popovers
and boxes is nothing has changed. On a 256 node lan, this drives one
core almost to 100% each time. Well, David's cores. My cores barely
notice it. :-)

end so that everything can run in parallel. And, so the extend
button (among others) is not locked out from working while it loads
everything.

Also, do not copy the massive idledata strings through the database
(twice). Sometimes a file is just a good idea.

(Verification continues to be off by default, as before; caller may not
have the cacert.  However, if you specify --verify on an Emulab server
or client, it will check the canonical cacert locations for you if you
don't specify via --cacert.)

experiment creator, so that the XMLRPC interface can be used. I deem
this safe in the geni world.

transient dropouts on the wireless.

script. Also add image relocation support for aggregates that do not
store images created there.

image relocation.

so that the scheduler will retry failed aggregates.

resolved.

Specifically, a "node" name cannot contain a colon and "port" and
"card" must be decimal digits.

nodes causes Jacks to hang the page for minutes. The threshold is now
75 nodes until Jacks is fixed.

Not sure I like this approach, might revisit later.

* Add feature check that allows users to use aggregates. Used so that
  not all users see the FE nodes. The same feature checks are used on
  the resource availability page.

* Turn cluster selector on.

* Mess with ordering, since Emulab is not a federate in Powder,
  everything else is, and we want Emulab first and Cloudlab Utah second,
  and everything else after that. Silly, but I know this will change so
  using brute force approach for now.

"radios" at the moment, so other cluster get a single wide graph.

some confirmation before powering down all of Powder.

using -r to remove CA certificate.

where they are used, not to all aggregates in a multisite experiment.