Apply Thomas Vachuska's solution of ONOS-537 (#2177) to FlowModBuilderVer13.
This solves the exception mentioned in the comment of ONOS-540.
Tested with OpenvSwitch 2.3.0 in Mininet 2.2.0.
Change-Id: I1f9d27d1b978ae01d16d63d6377046e4e2f3cd00
This fixes a problem where a large number of incoming STATS messages
is practically using all available threds from the pool (16), and
there are no available threads to handle the BARRIER messages.
Change-Id: I1130eb8f3b5a17d5d3a7825f32da68eacb99569a
fixing other threadpool issues, ie. not using cachedThreadPool
Change-Id: I40ef10e1f704aef779b2a23c0497dfb7992520eb
- Add current InstallationFuture information on TimeoutException
- Set timeout values propotional to batch size
- Fix for ConcurrentModificationException
- Check if src/dst is part of the graph before path computation
Change-Id: Iabeac7939c52502b83bf9ebcbe2023539de3ae99
The handling is not actually implemented, I just want to suppress the log
message for now. I have a task on Jira to actually implement this: ONOS-333
Change-Id: I9b230e7c602a61f3c2a0c35ae04ae06e305cd439
Added ids to Flow batch futures.
Adding some basic unit tests for IntentManger
Adding failedIds to the completedOperation in FlowRuleManager
Change-Id: I7645cead193299f70d319d254cd1e82d96909e7b
Fixed some code that was reporting findbugs errors
Implemented a suppression mechanism for findbugs
reported errors and a suppression file.
Change-Id: Ie8a2e84cc57ec6ddfa14d782ee89feb095b9dd59
Problem should now be fixed. Hashing on enums last is a bad
idea because the enum value could be 0.
Change-Id: Ib29e90b393b5285be2807729b52e69b121340f09
and FlowModBuilderVer13) to use Ip4Address and Ip4Prefix
instead of the generic IpAddress and IpPrefix when handling
IPv4 source and destination.
Change-Id: If35d735efe4cf557c79dc04eccf38f6a909e4506
- roleChanged(), isReachable()
Eventually consistent nature of Device store was
interfering with mastership control.
Change-Id: I9c0dd846a4e30863f922f6706c6cb62fd7c83f29
- For some of the methods, the IP Version needs to be specified by the
caller. This applies to the IpPrefix API as well.
- For now, everywhere IpAddress is used and the IP version has to be
explicitly specified by the caller, we assume/specify IPv4.
- Added unit test for class IpAddress: for both IPv4 and IPv6 addresses.
* Now IpPrefix uses IpAddress to represent the subnet address
* The IpPrefix subnet address is masked-out by the prefix length.
E.g., IpPrefix("1.2.3.4/24") is now stored as IpPrefix("1.2.3.0/24")
* Removed IpPrefix methods that are not used or don't apply anymore
* Replaced usage of IpPrefix with IpAddress where appropriate
* Removed IpAddress.MAX_INET_MASK and use IpPrefix.MAX_INET_MASK_LENGTH
instead
* Renamed IpAddress.INET_LEN to INET_BYTE_LENGTH
* Added IpAddress.INET_BIT_LENGTH, INET6_BYTE_LENGTH, and INET6_BIT_LENGTH
* Removed methods that are semantically incorrect, or are not needed/used
- IpAddress.prefixLength()
- IpAddress.toPrefix()
- IpAddress.mask()
- IpAddress.netmask()
- IpAddress.network()
- IpAddress.host()
- IpAddress.isMasked()
- IpAddress.contains()
- IpPrefix constructor for version and bytes (but no netmask)
- IpPrefix.valueOf(int)
* Misc. other cleanup.
