[
https://issues.apache.org/jira/browse/CASSANDRA-7620?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14077870#comment-14077870
]
Benedict commented on CASSANDRA-7620:
-------------------------------------
No, you won't get deterministic results here, as there's no way to generate
consistent random numbers for the selection operations with concurrent clients.
You need your insert part of the yaml to be:
{code}
insert:
partitions: fixed(1) # number of unique partitions to update in a
single operation
# if perbatch < 1, multiple batches will be
used but all partitions will
# occur in all batches (unless already
finished); only the row counts will vary
pervisit: fixed(1)/1 # ratio of rows each partition should update
in a single visit to the partition,
# as a proportion of the total possible for
the partition
perbatch: fixed(1)/1 # number of rows each partition should update
in a single batch statement,
# as a proportion of the proportion we are
inserting this visit
# (i.e. compounds with (and capped by)
pervisit)
batchtype: UNLOGGED # type of batch to use
{code}
> Allow user stress commands to specify the seed on the command line
> ------------------------------------------------------------------
>
> Key: CASSANDRA-7620
> URL: https://issues.apache.org/jira/browse/CASSANDRA-7620
> Project: Cassandra
> Issue Type: Improvement
> Components: Tools
> Reporter: Russell Alexander Spitzer
> Assignee: Russell Alexander Spitzer
> Priority: Trivial
> Attachments: CASSANDRA-7620-2.1.patch
>
>
> Although I don't want to add to parameter bloat, it would be really useful if
> the seed for a particular profile could be set on the command line. This
> makes it easier to run the same command with many different seed settings
> without modifying the profile file.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
