Build: #1022 was successful Scheduled with changes by 3 people

Build result summary

Details

Completed
Queue duration
5 minutes
Duration
1 minute
Labels
None
Revision
0e84be7d1b86b8b2acff6da2b0215f6f2352b591
Total tests
466
Successful since
#922 ()

Tests

Code commits

Author Commit Message Commit date
Guus der Kinderen Guus der Kinderen 47ecde13e5c65284b396eca9c2ee2a26ce5aa3ab Fix wording of alert
Guus der Kinderen Guus der Kinderen 0e84be7d1b86b8b2acff6da2b0215f6f2352b591 OF-2313: Do not depend on clustered cache when resolving cluster outage
Code that facilitates working with incoming server sessions previously depended on two caches. The 'second' cache stored 'validatedDomains' information. This comment was added:

> Content is stored in a clustered cache so that even in the case of the node hosting the sessions is lost we can still have access to this info to be able to perform proper clean up logic

We now know that in case of an incident that involves losing a cluster node, the content of caches cannot be depended on (instead, local copies of cached data needs to be maintained).

This commit removes the second cache, replacing it with a construct that keeps 'validatedDomains' info in both the primary cache, as well as locally-stored data.
Guus der Kinderen Guus der Kinderen f4af42ccc4ca926f2ad73dde3d7d1853bc9f4a81 OF-2311: Remove resumable state when closing remote sessions
When a cluster node closes a (client) session that lives on a different cluster node, it's is very likely that this session is never going to be resumed (XEP-0198): the other node presumably has a good reason to want to close it (eg: replacing it with another session in IQBindHandler).

This commit ensures that a session on a different cluster node does not fall in a 'resumable' state when it is closed by another cluster node.
Guus der Kinderen Guus der Kinderen 69686dc594287d5f1d6cba85cf0258ffb323c859 OF-2300: Detect and recover 'lost' cache data after cluster breakage
When a cluster node drops out of a cluster, it might be unable to retrieve cache entries. This means that those cache entries are effectively lost.

To work around this, relevant caches are supported by local data structures that can be used to re-populate these caches.

This commit detects and recovers entries that is lost in caches (as described above) for all caches in SessionManager and RoutingTableImpl
Guus der Kinderen Guus der Kinderen ad4533eda5294b0c7cdd13942deca84d08894d0d Replaced term 'hostname' with 'domain' where appropriate in SessionManager
Older versions of Openfire used the term 'hostname' and 'domain' interchangebly. With clustering though, a single XMPP domain can consists of multiple servers, that each have their own hostname.

This commit replaces most references to the word 'hostname' with 'domain' in the implementation of SessionManager. No functional changes are intended.

Jira issues

IssueDescriptionStatus
10 more issues…
Unknown Issue TypeOF-2293Could not obtain issue details from Jira
Unknown Issue TypeOF-2294Could not obtain issue details from Jira
Unknown Issue TypeOF-2296Could not obtain issue details from Jira
Unknown Issue TypeOF-2297Could not obtain issue details from Jira
Unknown Issue TypeOF-2298Could not obtain issue details from Jira