GitLab

Add new script to "deprecate" images:

	boss> wap deprecate_image
	Usage: deprecate_image [-e|-w] <image> [warning message to users]
	Options:
	       -e     Use of image is an error; default is warning
	       -w     Use of image is a warning

When an image is deprecated with just warnings, new classic experiments
generate warnings in the output. Swapping in an experiment also
generates warnings in the output, but also sends email to the user.
When the image set for error, both new experiment and swapin will fail
with prejudice.

Same deal on the Geni path; we generate warnings/errors and send email.
Errors are reflected back in the Portal interface.

At the moment the image server knows nothing about deprecated images, so
the Portal constraint checker will not be bothered nor tell the user
until later when the cluster throws an error. As a result, when we
deprecate an image, we need to do it on all clusters. Needs to think
about this a bit more.

Mysql 5.7 refuses to load it since it tries to load NULL in a non-NULL field.
Apparently mysql 5.5 was okay with this.

this used to work, odd that we even did this.

(But then I just if-0'ed out the whole planetlab related query).

https://dev.mysql.com/doc/refman/5.7/en/sql-mode.html#sqlmode_only_full_group_by

partition/partitions/stored/virtual

Closes #251.

Had to add the "d820-explicit" desire.

(This check was added for rflinks, and it is necessary for them, but it
doesn't apply to all nodes.  So if the node it checks isn't actually in
the solution, don't check.)

We set this attribute for the Moonshot cartridges so that "power cycle"
will force a "power on" if it detects the node is off when attempting
a cycle operation.

Also, implement "power status" for the Moonshot nodes.

When spitting out an advertisement and the default osid is an
image alias, be sure to stick the alias name into the advertisement,
not the first of the images the alias points to. This is wht we
were getting -X86 in Jacks from the moonshot cluster.

1. Fix up multi architecture code wrt auxtypes (say, pvcm). I was
   failing to treat "d710-vm" as a node with a specified type, and so
   were doing multiple mapper runs with different architectures. This is
   cause an auxtype is not really a first class type object, but it is
   now, enough to recognize it as such.

   Aside; vtypes are still a potential problem, need to give this more
   thought.

2. Use the reservation MaxSwapInMap() to determine if the request is
   asking for more nodes of a specific type then allowed; do not bother
   to run assign, bail right away. We can do this we now track how many
   nodes of a specific type have been requested, for the multi
   architecture stuff. So might as well put to another good use.
   This close issue #236.

3. Add reservation system information dump to resinfo.log for post
   mortem debugging.

when checking the debug value throughout the file.

Breakout ports are identified with a ':\d+' suffix in their port
description. This extension of the existing hack for submodules lets
breakout ports be referenced by their ifindex instead of trying to
encode the combination of module, submodule, port, and subport into a
single port number in the database.  This is total crap, and makes me
hate the mod/port DB encoding we have even more.  It is on my TODO list
to move snmpit to using device-specific port identification based on
port description strings instead.

NOTE: The module number semantics have changed for the submod/subport
hack.  Instead of using whatever module the port is on (plus one...),
the module is set to '0'. This is what should be used in the 'wires' DB
table.

Currently this implies raw (-r) and not zeroing (!-z) because those
options touch the DB counter tables and I didn't want to modify
those to make them bigints.

generate a top level element in the emulab namespace that lists the
nodes that cat be reserved (physical node types with at least one
actual node in the database). Change portal_monitor to look for and
use this element when processing the advertisement.

Get rid of ELVIN_COMPAT and CONFIG_OPSVM from elabinelab land.
These options still exist throughout the install code, didn't touch that.

Add a couple of variables at the top for the server/node hardware types
and server OS image.

Fix all the elabinelab_X(Y) references. Turns out that:
    elabinelab_fixnodes("boss")
(with quotes) is not the same array element as:
    elabinelab_fixnodes(boss)
(without quotes). And do to the way we coded things, only
elabinelab_X("boss") (or "ops" or "fs") worked, elabinelab_X(boss) would
not. But, only elabinelab_X(mypc1) would work and not elabinelab_X("mypc1")!
Now you should be able to use the unquoted versions of boss/ops/fs/mypcN.
Fun times with TCL.

Let them change the type and style (but still not remove entirely).

Also, recode a section to avoid perl "jump into construct" warning.

MB written instead of bytes written.

1. Using frisbee events in libosload_new as a replacement for
   statically (hand waved) maxwait times for image loading. When frisbee
   is generating events, we use those to determined if progress is being
   made.

2. Convert the CM to using the libosload_new library directly (like
   os_setup does). This is conditional on the NewOsload feature being
   attached to the geniuser. Otherwise, we go through the old path.

Call me confused or call me a dope, time will tell.

that the saved stdin/stderr are closed and gone.

to revisit this when I have time.

(The original approach was reusing the existing experiment record when
trying to repeat admission control.  This did properly decrement the
"used nodes" count, so that reallocating the same node to the same
experiment wasn't counted twice; but unfortunately it didn't
effectively increase the "free nodes" count and so it was still
possible to encounter spurious failures.  This could affect both
swapmod and RedeemTicket.)