Currently we are trying to ensure that the parent
directory exists as a key. But it is really a directory
and etcd expects it to be a directory. So made the
change to ensure that the parent key is created as
a directory and not as a simple key.
Signed-off-by: Jana Radhakrishnan <mrjana@docker.com>
Currently there are 3 distinct operations performed by
datastore
- Pushing the data to the store
- Updating the Index of the local object
- Updating the cache (in case of localscope)
Without a lock racing datastore api calls can interleave
in various surprising ways. Best thing is to keep these
3 above operation inseparable. Use a datastore lock to
achieve this.
Signed-off-by: Jana Radhakrishnan <mrjana@docker.com>
Since libnetwork is going to provide createNetwork
notifications only once when the network is created
bridge network needs to save it's network state in
persistent store so that it becomes available even
after restart.
Signed-off-by: Jana Radhakrishnan <mrjana@docker.com>
Now that libkv supports concurrent access to boltdb, there is no point
in depending on timeout mechanism
Signed-off-by: Madhu Venugopal <madhu@docker.com>
Currently every `NewDatastore` creates a brand new
libkv store handle. This change attempts to share
the libkv store handle across various datastore handles
which share the same scope configuration. This enables
libnetwork and drivers to have different datastore handle
based on the same configuration but share the same
underlying libkv store handle.
This is mandatory for boltdb libkv backend because no two
clients can get exclusive access to boltdb file at the same
time. For other backends it just avoids the overhead of having
too many backend client instances
Signed-off-by: Jana Radhakrishnan <mrjana@docker.com>
Add local scope store caching support as
well as do some refactoring to make it datastore
scope aware and manage scope specific config.
Signed-off-by: Jana Radhakrishnan <mrjana@docker.com>
This way we won't vendor test related functions in docker anymore.
It also moves netns related functions to a new ns package to be able to
call the ns init function in tests. I think this also helps with the
overall package isolation.
Signed-off-by: David Calavera <david.calavera@gmail.com>
In that commit, AtomicPutCreate takes previous = nil to Atomically create keys
that don't exist. We need a create operation that is atomic to prevent races
between multiple libnetworks creating the same object.
Previously, we just created new KVs with an index of 0 and wrote them to the
datastore. Consul accepts this behaviour and interprets index of 0 as
non-existing, but other data backends do no.
- Add Exists() to the KV interface. SetIndex() should also modify a KV so
that it exists.
- Call SetIndex() from within the GetObject() method on DataStore interface.
- This ensures objects have the updated values for exists and index.
- Add SetValue() to the KV interface. This allows implementers to define
their own method to marshall and unmarshall (as bitseq and allocator have).
- Update existing users of the DataStore (endpoint, network, bitseq,
allocator, ov_network) to new interfaces.
- Fix UTs.
* Removed network from being marshalled (it is part of the key anyways)
* Reworked the watch function to handle container-id on endpoints
* Included ContainerInfo to be marshalled which needs to be synchronized
* Resolved multiple race issues by introducing data locks
Signed-off-by: Madhu Venugopal <madhu@docker.com>
