Note: When go-grpc-prometheus was in the main $gopath (even at this
version) and everyone else was where they always were in vendor/ this
didn't build! It gave errors like:
have SendHeader("github.com/purpleidea/mgmt/vendor/google.golang.org/grpc/metadata".MD) error
want SendHeader("google.golang.org/grpc/metadata".MD) error
and I got frustrated. Putting it "next" to the other vendored deps seems
to have fixed this. Where are the golang docs that explain this
phenomenon?
This also requires golang 1.8+ as that is a requirement for etcd. It's
probably a reasonable thing for us too.
Note the older versions of etcd had some bugs with the concurrency
package and other things, so this is a necessary bump.
Graph changes from autogrouped -> not autogrouped or vice versa cause a
panic (or I assume a leak) because we compared the auto grouped graph to
the ungrouped one, which would cause an Exit on an unstarted Vertex.
This includes a test that seems to reliably reproduces the issue.
This allows the implementer of the GAPI to specify three parameters for
every Next message sent on the channel. The Fast parameter tells the
agent if it should do the pause quickly or if it should finish the
sequence. A quick pause means that it will cause a pause immediately
after the currently running resources finish, where as a slow (default)
pause will allow the wave of execution to finish. This is usually
preferred in scenarios where complex graphs are used where we want each
step to complete. The Exit parameter tells the engine to exit, and the
Err parameter tells the engine that an error occurred.
Since the pgraph graph can store arbitrary pointers, we don't need a
special method to create the vertices or edges as long as they implement
the String() string method. This cleans up the library and some of the
examples which I let rot previously.
The graph of dependencies in golang is a DAG, and as such doesn't allow
cycles. Clean up this lib so that it eventually doesn't import our
resources module or anything else which might want to import it.
This patch makes adjacency private, and adds a generalized key store to
the graph struct.
This puts the generation of the initial event into the Next method of
the GAPI. If it does not happen, then we will never get a graph. This is
important because this notifies the GAPI when we're actually ready to
try and generate a graph, rather than blocking on the Graph method if we
have a long compile for example.
This is also required for the etcd watch cleanup.
This causes a graph to actually stop processing part way through, even
if there are poke's that want to continue on. This is so that the user
experience of pressing ^C actually causes a shutdown without finishing
the graph execution. It might be preferred to have this be a user
defined setting at some point in the future, such as if the user presses
^C twice. As well, we might want to implement an interrupt API so that
individual resource execution can be asked to bail out early if
requested. This could happen on a third ^C press.
This adds a P/V style semaphore mechanism to the resource graph. This
enables the user to specify a number of "id:count" tags associated with
each resource which will reduce the parallelism of the CheckApply
operation to that maximum count.
This is particularly interesting because (assuming I'm not mistaken) the
implementation is dead-lock free assuming that no individual resource
permanently ever blocks during execution! I don't have a formal proof of
this, but I was able to convince myself on paper that it was the case.
An actual proof that N P/V counting semaphores in a DAG won't ever
dead-lock would be particularly welcome! Hint: the trick is to acquire
them in alphabetical order while respecting the DAG flow. Disclaimer,
this assumes that the lock count is always > 0 of course.
Improvements in the engine have uncovered some annoying race conditions
which would cause the engine to block between transitions. This is a
test which catches the most obvious file based ones.
This requires inotify to work in the test environment.
* Check and install libvirt with Homebrew
macOS does not have apt, dnf or yum. Add checking for homebrew for
installing libvirt.
* Use platform timeout for tests
* Add timeout detection to test/util.sh
* Use $timeout for shell test requiring timeout
Add owner which must be username or uid of the file owner, group which is
the group name or gid of the file, and mode which is the octal unix file
permissions.
Add separate implementation for Go 1.6 and lower.
The mgmt graph depends on state tracking to eliminate redundant pokes.
With the Watch loop now able to produce events quickly, it should no
longer play a part in determining the vertex state. This simplifies the
resource API as well!
This is a monster patch that splits out the yaml and puppet based graph
generation and pushes them behind a common API. In addition alternate
pluggable GAPI's can be easily added! The important side benefit is that
you can now write a custom GAPI for embedding mgmt!
This also includes some slight clean ups that I didn't find it worth
splitting into separate patches.
If you don't give your two host cluster enough time to "feel healthy",
it will generate an error if you do operations within five seconds. This
is a regression and the five seconds is also quite arbitrary. This is
detailed at: https://github.com/coreos/etcd/issues/6305
This seems to be a bit of a race condition, even with a 10s timer, so
this also disables the StrictReconfigCheck. Re-enable this as soon as
possible.
The MemberList() function which apparently looks at all of the endpoints
was blocking right after a member exited, because we still had a stale
list of endpoints, and while a member that exited safely would update
the endpoints lists, the other member would (unavoidably) race to get
that message, and so it might call the MemberList() with the now stale
endpoint list. This way we invalidate an endpoint we know to be gone
immediately.
This also adds a simple test case to catch this scenario.
