-
Notifications
You must be signed in to change notification settings - Fork 72
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Update pipe_verify_handoff_blocking to handle vnode shutdown #1093
base: riak/2.0
Are you sure you want to change the base?
Conversation
rpc or other processes will cause one or two persistent processes to start up during the test, but the number shoould not increase after that. Altered the test to check for massive increase in processes over a larger number of faults rather than an equality of process_count over a relatively small number of faults.
Still needs some work!
* Backed out the change to riak_core_handoff_sender_intercepts which had a modified intercept * Modified the handle_command intercept to ensure we don't count the vnode crashes that were happening before the fix * Reverted the test to the original, but check to make sure we can read all the entries, including the ones we added during handoff
…ive handoff a chance to complete.
update some yokozuna tests from work done in riak/2.0 branch for 2.0.…
(by not waiting for handoffs, just pending changes, as the old test did)
convergence outside of other tests.
* add new types for dialyzer * add test to reload schema and overwrite it on type change * test nested search (e.g. nested json objects) * test re-putting fields, include crdt map workaround, discussed in https://gist.github.com/sdebnath/36c235e042cb35db7d1f * increase aae concurrency for core_prop test
The verify_snmp_repl test reproduces the error condition (one cluster configured for realtime replication with two or more other clusters) and passes if and only if the riak_snmp_stat_poller module correctly handles this configuration. https://bashoeng.atlassian.net/browse/RIAK-1884 basho/riak_snmp#27
…ugh handle_command, to ensure we are testing the case statements in riak_kv_vnode:handle_command with write_once puts.
Add basic cluster converge test to try to capture issues with cluster convergence outside of other tests.
Merge 2.0 onto develop-2.2
@JeetKunDoug There's a comment around line 110 that seems to refer to the old behavior involving |
Gotta run out for a bit, but other than that one comment, it all looks good to me. Once that's fixed, you can consider this +1'd. |
c1ecc9f
to
5f6a47a
Compare
After further code review, realized that the race condition mentioned isn't really what was going on - fixed the bug in riak_pipe:queue_work_wait in basho/riak_pipe#106 (RIAK-1723) (RIAK-1830). This test should fail without that fix, and pass with it. |
# Conflicts: # tests/yz_core_properties_create_unload.erl
Merge 2.0 into develop 2.2
dependency cleanup
- update crdt cap and converge tests for hlls - temp rebar updates for riak-client; *will update w/ correct version* upon mergies
Update verify_aae test to verify throttle config
This can be disabled by setting the `giddyup_post_debug_logs` config value to `false` (it is `true` by default).
This pulls in the latest client code, as well as newer riak_pb stuff which is needed for the latest develop-2.2 Riak code.
Update client libraries to use develop-2.2
All the HLL stuff has been merged into the client develop-2.2 branches, so we can go ahead and update these dependencies to match that.
hll tests (against develop-2.2)
Upload debug logs to Giddyup
Somehow managed to pass on 2.0.7, but shouldn't have (and doesn't in 2.2.0).
Add back call to stop nodes before deleting index directories.
Wait for CRDT search results, rather than failing after the first attempt.
Use intercepts to provoke an otherwise-occasional error where a vnode has been transferred after the vnode proxy has gotten its pid but before the work can be sent. Will fail until fix to `riak_pipe_vnode:queue_work_wait` is merged. See basho/riak_pipe#106 for the fix.
6507c1d
to
54a5e05
Compare
Occasionally the test would fail with an error where a vnode was transferred after the vnode proxy had gotten its pid but before the work could be sent.