[SR-Users] db_cassandra status and plans

[Lucian Balaceanu]

Hi guys,

At 1&1 we actually haven't used db_cassandra, but were aware of some of 
its shortcomings (e.g. thrift interface).
Do you consider Cassandra is stable and easy enough to maintain so as to 
use it as usrloc back-end? Are you
also considering other alternatives?

Have a nice day,
Lucian

On 31.03.2016 14:09, Daniel-Constantin Mierla wrote:
> Hello Andreas,
>
> On 31/03/16 12:57, Andreas Granig wrote:
>> Hi Daniel,
>>
>> On 03/31/2016 12:45 PM, Daniel-Constantin Mierla wrote:
>>> I haven't used Cassandra at all, maybe some of the devs at 1&1 can share
>>> if they have any plans or usage stats for it, being the initial
>>> developers of the module.
>> Let's see if someone chimes in. Not sure if Anca is still involved, or
>> if/how it's actually used at 1&1 or anywhere else.
>>
>>> For me it is ok to introduce a new module as well, the old one can be
>>> removed afterwards if the new one overtakes it in features.
>> Makes sense, although I doubt someone is using it with recent Kamailio
>> versions due to the issues at least I encountered with it.
>> The new module in itself will be a drop-in replacement of the current
>> one (so we could re-use the same name but rewriting it from scratch),
>> the only limitation I can see for now is that it won't support cassandra
>> version older than 1.2 due to lack of cql support. So having a grace
>> period (or a parallel module with a different name) is fine with me.
> Perhaps is good to just replace it with a fresh implementation. I don't
> think there will be much (any) use of the old version. Anyhow, the old
> module can be moved to obsolete folder and if someone needs to work with
> an older lib, it can be rescue it from there.
>
> Cheers,
> Daniel
>
>>> What I actually wanted to discuss is regarding the query for NAT pinging
>>> -- with recent versions, there should be a dedicated column for querying
>>> natted records. It is no longer relying of flags (and bitwise
>>> operations) for this case. Or did you have in mind something else with:
>>>
>>> """
>>> The module can not be used together with nathelper for nat pings, since
>>> the queries don't provide a key, which is required for cassandra.
>>> """
>> Actually one of the issues with the cassandra schema is that the new
>> columns introduced in location e.g. for natping are not reflected in the
>> module's table schema. I haven't looked into the details of the db query
>> the module tries to do in case of natping, I was just observing the
>> errors the module was throwing for each natping attempt.
>>
>> The plan is to use the new way of querying records for natping anyways.
>>
>> Andreas
>>
>> _______________________________________________
>> SIP Express Router (SER) and Kamailio (OpenSER) - sr-users mailing list
>> sr-users at lists.sip-router.org
>> http://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-users