Ignoring pr value from bucket type. about riak_kv HOT 6 CLOSED

The issue is on line 313 in riak_kv_get_fsm. The DEFAULT_PR is 0 instead of default.

To reproduce: have 3 riak nodes.
Test bucket properties:

test_pw is active

young_vclock: 20
w: 3
sync_on_write: backend
small_vclock: 50
rw: quorum
r: 3
pw: 3
precommit: []
pr: 3
postcommit: []
old_vclock: 86400
notfound_ok: true
node_confirms: 0
n_val: 3
linkfun: {modfun,riak_kv_wm_link_walker,mapreduce_linkfun}
last_write_wins: false
dw: quorum
dvv_enabled: true
chash_keyfun: {riak_core_util,chash_std_keyfun}
big_vclock: 50
basic_quorum: false
allow_mult: true
active: true
claimant: '[email protected]'

Stop one riak node.

Do a get:
I expect this to return with an pr_val_unsatisfied error:

 riakc_pb_socket:get(poolboy:checkout(riakc_pool), {<<"test_pw">>, <<"test">>}, <<"fooo1">>).
{error,notfound}

riakc_pb_socket:get(poolboy:checkout(riakc_pool), {<<"test_pw">>, <<"test">>}, <<"fooo1">>, [{pr, 3}]).
{error,<<"{pr_val_unsatisfied,3,2}">>}

from riak_kv.

Thanks, I see it now.

I've done a quick check of the PUT fsm as well as the GET fsm, and it looks like it is only this value where the error exists.

I'm testing some leveled changes at the moment which may require a 3.0.12 release, so I will bundle a fix for this in as well. Target is to have 3.0.12 with this fix by the end of the week.

from riak_kv.

Thanks, I could issue a pull request if it helps.

from riak_kv.

Is this specifically about setting in bucket types? I know there is a test for the 'PR' value here, but it is passed in as a GET option - https://github.com/basho/riak_test/blob/develop-3.0/tests/pr_pw.erl

In theory the bucket properties should be checked here, and then that is added to the get_core. The get_core checks, and only gives a positive response when satisfied.

This shouldn't change however PR is set. Do you have an idea where this is breaking down?

from riak_kv.

If you have time, that would help a great deal.

from riak_kv.

Included in Riak 3.0.12

from riak_kv.