A back poke is the deferral or delay of a Process/CheckApply. This is
because we notice that we're not truly ready to CheckApply due to some
timestamp issue. When Process errors, we should accept that, but not
treat it as a success.
The retry and limit "satellite" event loops didn't allow pausing or
resuming, and instead you needed to wait until either was done before
you could pause.
The downside of this patch is that for very fast graph transitions, we
wouldn't be really obeying the limits anymore, however now that we have
per resource kind+name uid, we can persist the limits across graph swaps
if we want to.
Most importantly, this allows us to exit entirely when we're stuck in
one of these satellite loops.
This adds a meta state store that is preserved between graph switches if
the kind and name match. This is useful so that rapid graph changes
don't necessarily reset their retry count if they've only changed one
resource field.
This simplifies the pause mechanism and also avoids a deadlock on error.
If the Worker shuts down completely, but before we've been removed from
the graph, then an attempted pause would deadlock if we didn't have an
escape hatch here.
This removes the unnecessary ack mechanism now that we have a
synchronous channel send to represent the pausing, rather than an
asynchronous channel closing.
There's always the fear that there is either a panic or a deadlock in
the highly concurrent engine resource code. I have not seen one recently
and I've been running some pretty concurrent tests. In the meantime, and
with my hopefully improved knowledge of concurrency, I decided to
rewrite some of the "uglier" parts of the engine. I think it is a lot
clearer now, and much less likely that there is a concurrency issue.
This has been tested by running the examples/lang/fastcount.mcl example.
There were some bugs about setting resource fields that were structs
with various fields. This makes things more strict and correct. Now we
check for duplicate field names earlier (duplicates due to identical
aliases) and we also don't try and set private fields, or incorrectly
set partial structs.
Most interestingly, this also cleans up all of the resources and ensures
that each one has nicer docs and a clear struct tag for fields that we
want to use in mcl. These are mandatory now, and if you're missing the
tag, then we will ignore the field.
Hetzner cloud resource using hcloud-go. Requires polling via Meta:poll param. This first commit provides a stable vm resource with support for the basic functions of creating, deleting and updating a live server instance. SSH key handling does still require some attention to make sure checkapply can detect and update live changes to the specified keylist. A dedicated hetzner:sshkeys resource might be in order to make sure the keyset is handled correctly if there are multiple hetzner:vm resources running under the same Hetzner project. All remarks for future improvements are indicated with a TODO prefix
Unexported fields should be ignored when mapping structs from Golang to
mcl, as this avoids issues where certain structs cannot be used in mcl
representation due to their structure that may be incompatible with
conversion. Disallowing private fields allows those to still be included
in the struct but ignored when mapping the types, allowing the mcl
representation of the type to stil be valid
Resolves: https://github.com/purpleidea/mgmt/issues/684
Signed-off-by: Joe Groocock <me@frebib.net>
There were a bunch of packages that weren't well documented. With the
recent split up of the lang package, I figured it would be more helpful
for new contributors who want to learn the structure of the project.
The old system with vendor/ and git submodules worked great,
unfortunately FUD around git submodules seemed to scare people away and
golang moved to a go.mod system that adds a new lock file format instead
of using the built-in git version. It's now almost impossible to use
modern golang without this, so we've switched.
So much for the golang compatibility promise-- turns out it doesn't
apply to the useful parts that I actually care about like this.
Thanks to frebib for his incredibly valuable contributions to this
patch. This snide commit message is mine alone.
This patch also mixes in some changes due to legacy golang as we've also
bumped the minimum version to 1.16 in the docs and tests.
Lastly, we had to disable some tests and fix up a few other misc things
to get this passing. We've definitely hot bugs in the go.mod system, and
our Makefile tries to workaround those.
Since we'll want to use them elsewhere, we should make these helper
functions. It also makes the code look a lot neater. Unfortunately, it
adds a bit more indirection, but this isn't a critical flaw here.
docker/client.NewClient() is deprecated in favour of NewClientWithOpts()
which takes a series of client.Opt functions to configure the API
client. As mgmt only passes the API version through, this simplifies the
NewClient() calls.
Signed-off-by: Joe Groocock <me@frebib.net>
These dependencies are maintained because the upstream repos bundle
vendor directories into the repos, which cause namespacing issues during
build. Git submodules don't strip the vendor directory whereas most
vendoring tools would.
Signed-off-by: Joe Groocock <me@frebib.net>
This constant value is strongly tied to the language, and little to do
with the engine. Move the definition into the lang/types package to
prevent circular imports between lang/types and engine/util.
Signed-off-by: Joe Groocock <me@frebib.net>
