In turn the DeviceManager checks the role and sends the proper one down to the provider to be applied.
Change-Id: Icab9ac30b6e269d8acf9d14896c9858beaedb6da
Also reduces default pipeconf watchdog probe interval
Also fixes NPE on GDP when device is unreachable
Change-Id: Ie2fe1874b0883a037596d9a555a2f8cc030a55a6
(cherry picked from commit d797e2c28505fbdbb8597038ebcf977a053bae72)
This patch is an attempt to solve issues observed when restarting both
switches and ONOS nodes. Most of the issues seemed to depend on a
brittle mastership handling when deploying the pipeline.
With this patch, GDP registers devices to the core with available=false
(i.e. offline) and marks them online only when the P4 pipeline has been
deployed to the device. A new PiPipeconfWatchdogService takes care of
deploying pipelines and producing event when devices are ready.
Moreover, we fix a race condition where pipeconf-related behaviors
were not found. This was caused by GDP enforcing the merged
driver name in the network config, while external entities (e.g.
Mininet) were pushing a JSON blob with the base driver name. This patch
removes the need to rely on such a trick and instead uses
pipeconf-aware logic directly in the driver manager (change #19622).
Finally, we fix issues in P4RuntimeClientImpl that were causing the
stream channel not detecting unreachable devices. The solution is to
follow gRPC APIs and re-instantiate a new channel once the first fails.
Change-Id: I6fbc91859c0fb58a6db3bc197b7081a8fe9f97f7
Refactoring aims at providing a more robust connection procedure of
P4Runtime-based devices.
Refactoring requires devices to be registered to the core with
available=false (i.e. offline), and marked online only when the P4
pipeline has been deployed to the device. Also, it leaves the duty of
handling pipeconf-specific drivers to the driver subsystem, instead of
having the GeneralDeviceProvider take care of it, which resulted to be
brittle.
Changes include:
- Choose not to mark device online when re-asserting mastership role by
declaring special device annotation.
- Lookup drivers in device projectable model unsing driver service
device-based method, which is consistent with the way other subsystems
look for device drivers.
- In DriverManager, when looking for device drivers, consider the case
of pipeconf-programmable devices, in which case a special "merged"
driver needs to instantiated at runtime.
- In PacketManager, push flow objectives only if device is available
(i.e. after pipeline has been deployed).
Change-Id: I80f78a4f40730a06cb82fa55fefab08a91a89268
- Refactor cluster.json to support internal/external nodes ('controller' and 'storage')
- Bootstrap embedded partitions when 'storage' nodes not present
- Update onos-gen-config script to generate cluster.json based on environment variables
- Update setup scenario to ignore missing $OCC# environment variables
Change-Id: Ia93b64e13d7a7c35ed712da4c681425e3ccf9fe9
* Upgrade Raft primitives to Atomix 3.0
* Replace cluster store and messaging implementations with Atomix cluster management/messaging
* Add test scripts for installing/starting Atomix cluster
* Replace core primitives with Atomix primitives.
Change-Id: I7623653c81292a34f21b01f5f38ca11b5ef15cad
Clone to CPU is available only for packets processed via multicast
groups. Can be changed in the future when implementation for clone
session APIs is available in PI and P4 targets.
Also:
- compile "fabric-full" profile and generate constants from it
- use interpreter to map logical ports to data plane port IDs
Change-Id: I7db30c08dcf69ed9c870748cce8a797bbd5d6f78
This device type is now advertized by the server
device driver. Also, the ONOS UI maps this new device
type to a glyph.
Change-Id: Ib4147676474b43202bbdff595a0fa0520b70fe91
Signed-off-by: Georgios Katsikas <katsikas.gp@gmail.com>
Most notably, we fix a bug in which some nodes were not able to find
pipeconf-specific behaviors for a given device. The problem is not
completelly solved but it's mitigated.
There's a race condition caused by the fact that the GDP updates the cfg
with the merged driver name before advertising the device to the core.
Some nodes might receive the cfg update after the device has been
advertised. We mitigate the problem by performing the pipeline deploy
(slow operation) after the cfg update, giving more time for nodes
to catch up. Perhaps we should listen for cfg update events before
advertising the device to the core?
Also:
- NPE when getting P4Runtime client
- Detect if a base driver is already merged in pipeconf manager
- Longer timeouts in P4Runtime driver and protocol (for slow networks)
- Configurable timeout in P4Runtime driver and GDP
- NPE when adding/removing device agent listeners in P4Rtunime handshaker
- Various exceptions due to race conditions in GDP when disconnecting
devices (by serializing disconnect tasks per device)
- NPE when cancelling polling tasks in GDP
- Refactored PipeconfService to distinguish between driver merge,
pipeconf map update, and cfg update (now performed in the GDP)
- Fixed PipeconfManagerTest, not testing driver behaviours
- Use Guava striped locks when possible (more memory-efficient than maps,
and with strict atomicity guarantees w.r.t. to caches).
Change-Id: I30f3887541ba0fd44439a86885e9821ac565b64c
The P4Runtime client was hanging (deadlock) on a master arbitration
request. As such, all other requests (e.g. table write) were waiting
for the client's request lock to become available.
Apart from fixing those deadlocks, this patch brings a number of
improvements that all together allow to run networks of 100+ P4Runtime
devices on a single ONOS instance (before only ~20 devices)
Includes:
- Asynchrounous mastership handling in DevicHandshaker (as defined in
the P4Runtime and OpenFlow spec)
- Refactored arbitration handling in the P4RuntimeClient
to be consistent with the P4Runtime spec
- Report suspect deadlocks in P4RuntimeClientImpl
- Exploit write errors in P4RuntimeClient to quickly report
channel/mastership errors to upper layers
- Complete all futures with deadlines in P4Runtime driver
- Dump all tables in one request
- Re-purposed ChannelEvent to DeviceAgentEvent to carry also mastership
response events
- Fixed IntelliJ warnings
- Various code and log clean-ups
Change-Id: I9376793a9fe69d8eddf7e8ac2ef0ee4c14fbd198
Two main changes for the bug-fix:
- avoid doing a full-reroute in the event of link failure as this will leave behind stale state in stores
if event has been preceded by mastership change leading to the nuking of the ecmpSpg for one of the
link's devices; instead do a rehash
- when full-reroute is attempted, do it only once, with a complete nuke of the next-obj store
Improvement in group handling allows for a max number of retries for a group that failed to be added.
Earlier behavior was to try only once, and if it fails, it gets removed from the group-store. Now it
is removed after a couple of retries - so total 3 attempts to program the group.
Change-Id: I54ca8203cef779463522b01353540d12f8be3c91
This patch gives us the ability to disable
host learning in certain ports via the ports
configuration.
We have the following cases :
- Discover a host with no learning configuration, this
results to learned host (missing config assumes learning is on)
- Discover a host with learning configuration set to false,
do not learn that host.
- Update the learning configuration of a CP to false. Then,
fetch all host at that location, if these hosts are not
configured statically at this location, remove them
Change-Id: I2a05ce5f9a890bb4049fd7856f95d78f467e3330
Includes:
- Use new P4Runtime "v1" package names
- Removed VALID match
- New table entry priority spec (1 is min priority, not 0)
- Fixed p4c-bm2-ss to include arch flag
- Re-compiled P4 programs with more recent p4c (with updated p4info)
Change-Id: I05908f40eda0f0c755009268fd261fb8bcc9be35
- Implement probe retry
- Switch to typical core/provider design pattern for HostProbingService
and as a result decoupling the dependency between SR and HostLocationProvider
Change-Id: I33a15af580677ea376b421ac3e26f9821dcca844
