Nick Marino
7dc4a3cc0b
Fix bug in replication2 delete tests
...
This code works fine if replication2 is run as a standalone test, but
replication2 is also used as part of other bigger tests, and it can
fail. The problem is that keys 1-100 are not necessarily always written,
depending on how the test is being run. By using the 101-200 keys to
test deletions instead, we can avoid this failure case.
2016-05-31 16:07:27 -04:00
Nick Marino
bf14dfd13c
Improve namings in response to review comments
2016-05-06 15:56:18 -04:00
Nick Marino
8daebc1663
Add test code to verify deletes for realtime repl
2016-05-04 14:38:25 -04:00
Cuyler Jones
ab61070927
Fixed parenthetical spacing and a few other text spacing issues
2015-02-28 21:34:38 -07:00
Cuyler Jones
82d9d57a24
Fixed spacing to adhere to Basho guidelines.
2015-02-28 21:24:59 -07:00
Cuyler Jones
c6d27d5a99
Replaced start up process with build_clusters; Removed some lingering refactoring cruft.
2015-02-25 10:23:33 -08:00
Cuyler Jones
ef0581cefb
Code clean up and small node logging additions
2015-02-24 14:48:58 -08:00
Cuyler Jones
57fde811ef
Refactored Replication 2 into compartmentalized tests with documentation. Removed superfluous code.
2015-02-24 14:41:38 -08:00
Engel A. Sanchez
556cb7210c
Ensure riak_repl service is up across the board
...
Trying to use the repl features before newly started nodes have
riak_repl completely initialized leads to all sorts of nasty crashes and
noise. Frequently it makes fullsync stuck forever, which makes a lot of
the tests fail.
This also tweaks the AAE fullsync tests to remove assumptions about
failure stats when AAE transient errors occur. The behavior in the
handling of those errors has changed recently with the introduction of
soft exits.
2014-12-18 16:07:00 -05:00
Kelly McLaughlin
d82e65c633
Fix a race condition in replication2 test
...
Remove an assertion based on reading keys a single time after realtime
replication is re-enabled in the test. Instead just rely on the wait
condition that already followed the assertion to read and verify the
same keys.
2014-05-29 16:56:21 -06:00
Micah Warren
43ae4e2871
Removed 'allow_mult false' for test.
...
This reverses an earlier change to support a feature that has been stripped
(for now). When said feature is put back in, it should support multi.
Setting this to allow mutlt = true allows for more confidence in tests.
2014-05-19 12:48:08 -05:00
Jared Morrow
24619c638b
Merge pull request #513 from basho/bugfix/sdc/allow-mult-default
...
Fix tests that used the wrong name for the riak.conf entry for allow_mult
2014-01-24 09:54:17 -08:00
Christopher Meiklejohn
ba45350da3
Fix overlapping key ranges in replication test.
...
Fix overlapping key ranges which cause sibling generation because of
puts with no vclock. Add additional checking to assert this is true.
2014-01-21 20:22:17 -05:00
Sean Cribbs
59976f25f0
Fix typo true -> false
2014-01-19 12:00:21 -06:00
Sean Cribbs
34b2d3723c
Fix tests that used the wrong name for the riak.conf entry for allow_mult.
2014-01-19 11:47:50 -06:00
Jon Anderson
88fc799bb8
set allow_mult=false on test bucket to prevent crash in mutator; fix rhc errors
2013-11-15 07:47:32 -05:00
Dave Parfitt
f11ed61511
riak_core http setting moved to riak_api
...
use a generic wrapper to work with multiple versions of Riak
2013-09-26 18:00:10 -04:00
Andrew Thompson
f6d0e7fb6d
Verify the partition count is correct only on 1.4 and later
...
1.3 is buggy, so the check will always fail.
2013-07-03 13:57:33 -04:00
Chris Tilt
571e92d6ed
Test for fullsync's successful_exists status item
...
* After a fullsync, make sure that the numner of successful exists equals the number of partitions, since each partition should have one successful fullsync source process exit.
* We could extend this test to handle the parts of replication2 that have down nodes in the future.
* Added a utility function to get the number of partitions from a node, plus a function to get fullsync status items.
* Test passes with changes from branch cet-successful-exists-fix
2013-06-13 12:57:23 -07:00
Chris Tilt
2415895306
Merge pull request #282 from basho/pdx-aae-fullsync
...
Riak test for AAE fullsync
2013-05-22 15:28:26 -07:00
Chris Tilt
69ef5babff
Remove debug cruft.
2013-05-22 11:47:19 -07:00
Chris Tilt
3fd920efbb
clean up some stuff that got hacked while debugging and benchmarking
2013-05-15 20:19:48 -04:00
Joe DeVivo
08ac3fdbed
refactored config and local into separate modules
2013-04-26 08:24:58 -07:00
Chris Tilt
1a9bb3d883
AAE fullsync works, plus some new AAE r_t utils
2013-04-19 12:57:58 -07:00
Chris Tilt
65356efec1
Merge branch 'master' into cet-repl2-upgrade-test-194
2013-03-21 10:56:57 -07:00
Chris Tilt
9e0c249da1
Use rt functions for HTTP and restore lager:info for pb connection establishment
2013-03-20 10:31:44 -07:00
Dave Parfitt
823c65a871
use rt:pbc to read the new pb app.config format
2013-03-20 12:52:08 -04:00
Chris Tilt
950db199af
Improve Upgrade test for replication2
...
* Check for node stopped after asking it to stop in the upgrade
* Add better debugging to aid in determining which stage failed
2013-03-15 14:35:33 -07:00
Chris Tilt
ea39ada538
Export replication test entry points to share with new upgrade test
2013-03-15 14:32:23 -07:00
Joe DeVivo
679a5b1bd9
Added some waits in replication2 from @buddhisthead
2013-03-05 14:41:02 -07:00
Andrew Thompson
aa5f97cc92
Merge pull request #216 from basho/adt-wait-for-repl
...
Wait for repl to actually start before shutting the node down again
2013-02-21 09:55:23 -08:00
Andrew Thompson
bcc5a117d8
Wait for repl to actually start before shutting the node down again
2013-02-21 12:53:25 -05:00
Andrew Thompson
14136b21a3
Use an alphanumeric md5 hash for bucket to avoid angering HTTP
2013-02-21 12:35:57 -05:00
Dave Parfitt
90ba560cc6
Merge pull request #204 from basho/dip_shutdown_dw
...
change w -> dw for shutdown tests
2013-02-16 12:07:42 -08:00
Andrew Thompson
d7dca09df9
Make sure we exercise the code path when # of keys > diff_batch_size
...
We do this by lowering diff_batch_size to 10, and then writing 10,000
keys to a 64 vnode ring, thus ensuring that each vnode will have more
than 100 keys.
2013-02-15 14:03:31 -05:00
Dave Parfitt
1a67cb6a89
change w -> dw for shutdown tests
...
this is @vagabond's patch
2013-02-14 15:53:19 -05:00
Andrew Thompson
5af919b812
Make sure we wait for elections to complete when starting repl tests
2013-02-14 13:40:47 -05:00
Chris Tilt
d80c30025c
Remove export all from repl tests; make repl_util API explicit
2013-01-25 10:43:29 -08:00
Chris Tilt
1a61e06634
Moved common repl2 functions to repl_util and added log_to_nodes tracing to replication2_dirty test
2013-01-25 10:12:34 -08:00
Chris Tilt
b0ca6fa2c5
Move log_to_nodes into rt module for general use.
2013-01-24 15:51:33 -08:00
Chris Tilt
a076531785
Started refactor and added cluster-wide logging from tests.
2013-01-24 14:24:16 -08:00
Chris Tilt
cfb7ab29c1
Cleaned up changes per peer review. Replication test passes.
2013-01-23 17:09:27 -08:00
Chris Tilt
11ec2c3380
Added wait for ring to converge after bucket creation for both tests.
2013-01-23 15:55:04 -08:00
Dave Parfitt
9e56bf7fd3
repl test cleanup
2013-01-23 14:48:21 -05:00
Andrew Thompson
c14a337ed0
Merge pull request #128 from basho/adt-repl-bnw-shutdown
...
Add better 'writes while shutdown' tests for replication
2013-01-21 10:58:26 -08:00
Andrew Thompson
83ed717b42
Use multiple workers to write while shutting down
2012-12-17 16:36:17 -05:00
Engel A. Sanchez
0e8ecce7d5
Add riak_test behavior
2012-12-14 10:50:49 -05:00
Andrew Thompson
a720f2beda
Better write during shutdown tests
2012-12-09 00:51:11 -05:00
Andrew Thompson
65056cb12b
Hack in support for doing PB/HTTP writes while shutting down
2012-12-08 03:56:39 -05:00
Dave Parfitt
b07c0935ba
Merge branch 'dip_bnw_repl' of github.com:/basho/riak_test into dip_bnw_repl
2012-12-05 18:38:44 -05:00
