GitLab

We currrently have a few cases where a slice record exists, but
no sliver, and so Renew was failing. Since we store all of the
expiration in the slice record, we do not actually need to have
an aggregate, so remove the check.

Usage: setexpiration [-f] [-m <time> | -M] [-e <datetime> | -E] [-i | -I] <slice>
Options:
  -f      - Force operation even it makes no sense.
  -m      - Max increment time for a renew. In minutes.
            Use zero to allow anything. Use "null" to clear.
            Use "NN days" or "NN hours" also.
  -M      - Clear max increment time for a renew.
  -e      - Termination date; sliver may not be renewed past this
            date. Use standard date format (YYYY-MM-DD HH:MM:SS)
  -E      - Clear max termination date.
  -i      - Set 'idle ignore'; no idle checks or email.
  -I      - Turn idle checks back on.

to normal newnode mode.

The GPO purchased racks all look the same so we can add the new
nodes statically from the XML file that HP sends us.

Well, for the UEN rack and other privately purchased racks, we do not
have any of that info, and so we have to figure things out ourselves.
We could use the standard newnode route without modification, but that
is a realy pain, say on a 33 node rack where the nodes land in any
order and you have no idea what the corresponding ilo is.

So I whacked this script to run from a data file we generate, which
contains the two bits of info from the top of the node (ilo passwd and
the ilo hostname), and the IPs we want to assign to the node and the
corresponding ilo.

So after turning all the nodes on and they checkin as usual, we run
this script with the -r option, which looks at the dhcpd.lease file to
find out what IP is which ILO, and the it asks each ilo via XMLRPC for
its hosts daya, which includes the ethernet MACs. I use that to find
the corresponding node in the leases file. Now I reorder the new_nodes
table so that pc1 is at the bottom and pcNN is at the top.

Now go back to the web interface and add all of the nodes.

Then run this script again (no -r), which will create all of the
management interfaces in the DB, reset the ilo, and then do the
normal initilo stuff for each node (boot order, password change, ssh
keys, etc).

Presto, done.

I figure out how to get the switches to do ssh login, so not
you can just do "sudo ssh procurve1" instead of telnet.

from control node.

fumbles around in an infinite loop.

work fine when the nodes are behaving themselves.

1) geni_update_users: Takes a slice credential and a keys argument. Can
  only be invoked when the sliver is in the started/geni_ready state.
  Moves the slice to the geni_updating_users state until all of the
  nodes have completed the update, at which time the sliver moves back
  to started/geni_ready.

2) geni_updating_users_cancel: We can assume that some nodes will be whacky
  and will not perform the update when told to. This cancels the
  update and moves the sliver back to started/geni_ready.

A couple of notes:

* The current emulab node update time is about three minutes; the
  sliver is in this new state for that time and cannot be restarted or
  stopped. It can of course be deleted.

* Should we allow restart while in the updating phase? We could, but
  then I need more bookkeeping.

* Some nodes might not be running the watch dog, or might not even be
  an emulab image, so the operation will never end, not until
  canceled. I could add a timeout, but that will require a monitor or
  adding DB state to store the start time.

continual retries that annoy people.

Users want to make images derived from standard ones that support blockstores,
so let's make that possible.

Tweaks to prevent adding vlan 1 to the trunk allowed list.

We should only ever do this implicitly if the last vlan in the allowed list
is removed.

Things were confused.  A more explicit split and treatment is needed.

Update for portchannels in getPortState()

It gets called many times for a variety of commands...

* warnings normalized.
* Added logic to deal with empty allow list for trunks
* Added some missing locking.
* Added some comments.

Also added a new test in the test harness for trunks.

On to testing.

Still have a smill bit of logic to write dealing with adding/removing the
last vlan from the allowed list on "equal" trunks, and then it's on to
testing.

Based on snmpit_arista and snmpit_hp.

This module wraps the Mellanox XML-gateway API, an "interesting" sort of
interface offered up by Mellanox switches running MLNX-OS.  This module
handles all of the connection setup and XML encoding / decoding.  Users
of the module pass in arrays of commands to run, and arrays of results
are returned.  All commands in lists passed to a single call() invocation
will be passed along together to be invoked, in order, on the switch.
Mixtures of 'get', 'set-*', and 'action' commands are allowed.

This check-in also includes a test harness for this module.

Sample use:

my $gw = MLNX_XMLGateway->new("user:pass@switch.somehost.org");
my @cmds = (
  ["get", "/some/REST/path"],
  ["set-create", "/some/REST/path/to/create"],
  ["set-modify", "/some/REST/path/to/modify=newvalue"],
  ["action", "/some/REST/action/path", {var => value, ...}]
);

$results = eval { $gw->call(\@cmds) };

if ($@) {
  die "error: $@";
}

foreach my $res (@$results) {
  # path, type, value
  print "@$res\n";
}