Any comments?
It seems to run intermittently on my local VMs and just outright fails on a
Joyent Cloud or AWS image.
From: Kevin Burton [mailto:rkevinbur...@charter.net]
Sent: Tuesday, February 26, 2013 6:36 PM
To: 'Jeremiah Peschka'; 'riak-users'
Subject: RE: MapReduce performance probl
Any idea what happened? We have had to remove the riak db and started
from scratch to get rid of the ghost keys...
On Thu, 7 Mar 2013 11:35:12 +0100
Simon Effenberg wrote:
> Now we see only 3 occurrences of the keys. So maybe the reducing of the
> n_val could be a problem.. after we removed the
I understand. I was just suggesting different hardware. That's all.
wes
On Thu, Mar 7, 2013 at 2:20 PM, Paul Peregud wrote:
> It is not really a question of saving 55$ per disk - I'm wondering if such
> Riak cluster setup is feasible.
>
>
> On Thu, Mar 7, 2013 at 10:14 PM, Wes James wrote:
>
It is not really a question of saving 55$ per disk - I'm wondering if such
Riak cluster setup is feasible.
On Thu, Mar 7, 2013 at 10:14 PM, Wes James wrote:
> You can get crucial m4 512 for $395 on amazon.com. ocz are $450.
>
>
> http://www.amazon.com/Crucial-512GB-2-5-Inch-Solid-CT512M4SSD2/d
You can get crucial m4 512 for $395 on amazon.com. ocz are $450.
http://www.amazon.com/Crucial-512GB-2-5-Inch-Solid-CT512M4SSD2/dp/B004W2JL3Y
I don't know if you can get these out of the USA or not (you are out of
USA right?)
wes
On Thu, Mar 7, 2013 at 1:38 PM, Paul Peregud wrote:
> My fir
Thanks for the detailed explanation of what you're seeing. As it turns out,
your results are typical.
Usually the deletion happens automatically across all nodes, but as you've
seen, there's no guarantee of that. Reading a key will trigger read repair to
clean up lingering objects, and is effec
My first post to the list.
I want to store 20M of bounded FIFO queues (each queue has at most 50
items, 1KB each). Distribution: 80% of updates at 20% of queues.
Insertion rate: at most 5K per second, random; 50M per day.
Reading rate: at most 10K per second, each read reads all 50 items.
Once in
In our tests we are adding 3000 keys into 3-node Riak db right after nodes
have joined.
For each key one node reads it and modifies it and another node does the
same but also deletes the key when it sees other change (key is no longer
needed). After all keys are processed our test framework checks
Salut Godefroy
On Thu, Mar 7, 2013 at 6:50 AM, Godefroy de Compreignac
wrote:
> Hello,
>
> I'm running a cluster of 4 nodes (1,8 TB on each) and I have a problem of
> balancing. Current data repartition is 18%, 19%, 30%, 34%. The node with 34%
> of cluster data is completely full and doesn't want
hi,
i would like to ask if there is any timeline when this bug in "riak_search"
gets fixed, see pull requests:
https://github.com/basho/riak_search/pull/136
https://github.com/basho/riak_search/pull/137
for me it's no problem to apply the patches during installation of riak,
still it would be ni
Hello,
I'm running a cluster of 4 nodes (1,8 TB on each) and I have a problem of
balancing. Current data repartition is 18%, 19%, 30%, 34%. The node with
34% of cluster data is completely full and doesn't want to start anymoe.
I don't know what to do. Do you have a solution for such a problem?
Th
Now we see only 3 occurrences of the keys. So maybe the reducing of the
n_val could be a problem.. after we removed the keys (or tried it) they
were deleted and in the keys=keys output only 3 (was 4 before) are
listed. So somehow the increasing n_val from 3 to 12 (factor 4...) and
then reducing aga
Hi Mark,
we have 12 Riak nodes running. The exact command for getting keys is:
curl http://localhost:8098/buckets/config/keys?keys=true
Properties are:
curl -s http://localhost:8098/riak/config | python -mjson.tool
{
"props": {
"allow_mult": true,
"basic_quorum": false,
Hello,
sorry that I did not clarify that but I am talking about using Riak's links and
"Link walking".
I don't know how it implemented under the hood, but it's still special case of
MapReduce request, which could be (in most cases) slower than fetch by keys?
We are using have's http client but
This is only an idea:
You could mark your related objects with a 2i, when you need them
retrieve the list of IDs from that 2i, fetch them concurrently into some
local memory cache, and then navigate through your object graph by
fetching them one by one from your local cache. It will be hard to
Hello guys,
I have some question about using a links.
In our data model each object A has a lot of links to other objects and we are
doing a lot of queries to fetch all objects that linked with A.
Could it be a performance issue?
Will it be more faster to store list of ids of related objects
16 matches
Mail list logo
