GitLab

This is for the benefit of the cache cleaner so that it can find out
if an image file is still valid and/or up-to-date.

image_import by URL to get initial images.

admins see it.

size, which for the first time has bit us. I need to change the schema, but
for now make sure we truncate/uniqify project names longer then 32 chars.

First, a "bug fix": By coincidence the snmpit Cisco module was able to
build up the mod/port-to-ifindex mapping on newer switches where
interface descriptions include a submodule ID (e.g.
"GigabitEthernet1/0/1" vs. "GigabitEthernet1/1"). However, it was
actually just grabbing the submodule ID in place of the module ID.  This
is OK (in some loose sense) for non-modular switches where everything is
effectivley on the same module. Not good for IOS/NX-OS switches with
actual modules and submodule identifiers. Things would not have worked
for these.  As a fix that retains backward compatibility, the Cisco
snmpit module now correctly extracts the module ID, but subtracts '1'
from it.  This allows existing installations that have zero-based module
numbering in their database for non-modular IOS/NX-OS switches with
submodule IDs to continue to work as is.

The above nonsense is not what I set out to do, however, and the commit
just gets worse. I extended the hack for non-modular switches with a mix
of gigabit and ten gigabit to bump any ten gigabit interface port's
module ID to "1". The existing hack already did this for non-modular
switches with a mix of fast Ethernet and gigabit.  Absolutely horrific.
I need a shower.

Did this in the FreeBSD version previously.

returns an error, so we can retry. Just a bandaid, need to figure
out why pubsub is not reconnecting when the firewall idles the
connection out.

Includes password, so that link is gone from the actions menu.
Note that Portal users can change their email now, since we do
the email verification dance.

Reset on local/remote blockstores ensures that there is no blockstore related
state left in the root filesystem (e.g., mounts in /etc/fstab, iSCSI config,
LVM/ZFS state). It does this in such a way that upon reboot, all the necessary
state is recreated.

What this means is that you should now be able to take an image of a node
that uses blockstores and have that image actually work on another node!
Previously, there could/would be leftover blockstore turds that would make
the new image fail to boot.

Of course, this won't work until the standard images are remade and will
then only work for those images or images derived from them.

This is when we are cleaning up to make an image. Seems like the correct
thing to do is process the startup scripts in reverse order like we do for
a "shutdown". Mostly this doesn't matter, but it does matter for the
blockstore cleanup where we need to process the remote BSes before the locals.

We were always setting size, even for delta images. This will have to
be revisited when we start doing image directories where we might have
BOTH full and delta images.

event.

(from anything else).

to check the reservation_name in the nodes table; if reserved_pid is
set but reservation_name is null, it was prereserved via the web interface
not via the prereserve command line tool. In this case, we never clear
the pre-reserve, someone has to do that via the web interface.

It was checking the count of database rows (which would always have
been 1), not the count of free addresses.

deleted. I guess we should not allow users to do this, eh?

I'm way to frustrated at this point. Back to using master ...

When creating a pre-reserve, new -n option to specify a name for the
reservation, defaults to "default". All other operations require an
-n option to avoid messing with the wrong reservation. You are not allowed
to reuse a reservation name in a project, of course. Priorities are
probably more important now, we might want to change the default from 0 to
some thing higher, and change all the current priorities.

For bookkeeping, the nodes table now has a reservation_name slot that is
set with the reserved_pid. This allows us to revoke the nodes associated
with a specific reservation. Bonus feature is that when setting the
reserved_pid via the web interface, we leave the reservation_name null, so
those won't ever be revoked by the prereserve command line tool.

New feature; when revoking a pre-reserve, we now look to see if nodes being
revoked are free and can be assigned to other pre-reserves. We used to not
do anything, and so had to wait until that node was allocated and released
later, to see if it could move into a pre-reserve.

Also a change required by node specific reservations; when we free a node,
need to make sure we actually use that node, so have to cycle through all
reservations in priority order until it can used. We did not need to do
this before.

statewait) are not killed by the stateful firewall.

accounts/passwords). Thanks to Mike Blodgett for cluing us in to this and
providing the configuration he was using. This will run from cron a couple
of times a day. The setup is not quite fully automated yet, need to create
/usr/local/etc/medusa/{wordlist.txt,userlist.txt} by hand, as well as the
crontab entry.

We are scanning all local nodes (including VMs) as well as any allocated IP
addresses in address pools.

Minor tweaks so that ipfw/dummynet sets up properly. Allow dynamic
config (delay_setup) to accept a BW value up to 1Gb (was 100Mb).
Somewhat dubious: if the incoming interface speed matches the desired
shaping BW, set the BW to unlimited on the pipe (since the interface
is already throttling for us).

BEWARE! The fidelity of shaping at over 1Gb (e.g., adding delay to a
10Gb link) is awful. Even a 10Gb pipe with no delay or loss only achieved
about 7Gb/sec max. The delay node code needs serious work!