Hi Praveen,
There are a few things that could be contributing the to 404s.
The most likely issue has to do with your "n" and "r" values. With an
"r" of 1, and a laggy node in your cluster, you could have a situation
that resembled a netsplit. (At the moment, Riak does better with
downed nodes than laggy ones.) Laggy responses will result in you
returning a notfound to the client if you're only waiting for 1 of 2
existing replicas and have "{notfound_ok, true}".
Potential solutions to this (assuming it's the problem):
* consider retrying a number of times with your client logic; there is
some information about why you might want to do this on the wiki [0]
* change to "{notfound_ok, false}" because otherwise if a "not_found"
comes back on the first response, it'll be treated as valid
* increase R to 2.
There are a few other potential culprits, but this is the most-likely.
Let us know if any of that helps.
Mark
[0] http://wiki.basho.com/Client-Implementation-Guide.html
On Mon, Sep 17, 2012 at 4:01 AM, Praveen Baratam
<[email protected]> wrote:
> The frequency of error is now more common. Upto 1 failed request in 10. This
> is breaking everything.
>
>
> On Mon, Sep 17, 2012 at 3:04 PM, Praveen Baratam <[email protected]>
> wrote:
>>
>> Here are some more details about the cluster.
>>
>> {ring_creation_size, 1024},
>>
>> {default_bucket_props, [
>> {n_val, 2},
>> {r, 1},
>> {w, 1},
>> {allow_mult, false},
>> {last_write_wins, false},
>> {precommit, []},
>> {postcommit, []},
>> {chash_keyfun, {riak_core_util, chash_std_keyfun}},
>> {linkfun, {modfun, riak_kv_wm_link_walker, mapreduce_linkfun}}
>> ]},
>>
>> {storage_backend, riak_kv_multi_backend},
>> {multi_backend_default, <<"store">>},
>> {multi_backend, [
>> {<<"store">>, riak_kv_bitcask_backend, [
>> {data_root, "/var/lib/riak/bitcask/store"}
>> ]},
>> {<<"cache">>, riak_kv_bitcask_backend, [
>> {data_root, "/var/lib/riak/bitcask/cache"},
>> {expiry_secs, 21600}
>> ]}
>> ]},
>>
>> On using Riaknostic I am seeing these errors.
>>
>> [warning] The following preflists do not satisfy the n_val:
>> [[{973382926425464638881751031164555683013033197568,'[email protected]'},.....
>> a big list similar to this...]
>>
>> Please help. We are stuck.
>>
>> Praveen
>>
>>
>>
>> On Mon, Sep 17, 2012 at 4:18 AM, Praveen Baratam
>> <[email protected]> wrote:
>>>
>>> Hello Everybody,
>>>
>>> We are currently alpha testing out our Webapp which uses Riak for storing
>>> binary files (< 2 MB) with n_val as 2. We currently have 3 nodes in the
>>> cluster and all files are stored in the same bucket. We are using Bitcask
>>> engine.
>>>
>>> Sometimes we see 404 error cropping up when accessing these keys. This
>>> happens only once in a day or so and all existing keys are accessible most
>>> of the time. We are only trying to access the keys that we have put in Riak
>>> so there is no doubt of their existence.
>>>
>>> Can a network split cause such an error? Is there anything we can do to
>>> avoid such corner cases.
>>>
>>> Thank you
>>>
>>> Praveen
>>
>>
>
>
> _______________________________________________
> riak-users mailing list
> [email protected]
> http://lists.basho.com/mailman/listinfo/riak-users_lists.basho.com
>
_______________________________________________
riak-users mailing list
[email protected]
http://lists.basho.com/mailman/listinfo/riak-users_lists.basho.com
