[Lognorm] Any reason we should not support regex-based field-type?

[David Lang]

On Mon, 3 Nov 2014, singh.janmejay wrote:

> I think module-loading call having a flag to turn on/off would be better
> than build-time flag, purely because it allows one to build a package and
> use it on several boxes, while keeping the feature off when not required. A
> more fine grained control, in that sense.

I was thinking in terms of something in the config file. We need it to be 
something that the person configuring rsyslog will see so that they know they 
are doing something slow. Most users don't compile rsyslog so they won't ever 
see the compile time flag.

module loading time seems fine.

David Lang

> We can support both levels of enabling, but then a lot of people may be
> confused as to why it doesn't work when they turned it on while building.
>
> 2 levels also seems a little over-protective to me. Kinda like Windows "do
> you really want to delete this file" prompts.
>
> Does module-loading time flag sound ok?
>
> --
> Regards,
> Janmejay
>
> PS: Please blame the typos in this mail on my phone's uncivilized soft
> keyboard sporting it's not-so-smart-assist technology.
>
> On Nov 3, 2014 10:52 PM, "singh.janmejay" <singh.janmejay at gmail.com> wrote:
>
>> Hex was just an example. I meant when someone needs to roll out something
>> very quickly, and it's not supported in the build they have, it's useful to
>> be able to do it using regex in rulebase but still completely working in
>> terms of rules, rather than having to resort to a mix of rules and
>> exec-template calls with regexp based property extractors etc.
>>
>> It's still the same amount of work, if someone is trying to do something
>> unsupported by lognorm. So performance effects will still show, just not in
>> lognorm.
>>
>> This allows for flexibility and clean way of parsing logs at possibly even
>> slightly lesser cost then the former approach.
>>
>> We can support a switch-on flag in conf file(module loading call, may be)
>> which can be used to enable performance sensitive features.
>>
>> I personally think warning in logs would be enough though.
>>
>> --
>> Regards,
>> Janmejay
>>
>> PS: Please blame the typos in this mail on my phone's uncivilized soft
>> keyboard sporting it's not-so-smart-assist technology.
>>
>> On Nov 3, 2014 9:22 PM, "Champ Clark III" <cclark at quadrantsec.com> wrote:
>>
>>>
>>> -----BEGIN PGP SIGNED MESSAGE-----
>>> Hash: SHA1
>>>
>>> I think David means something like:
>>>
>>> ./configure --enable-preformance-pigs
>>>
>>> I'm not a huge fan of this for a couple of reasons.
>>>
>>> In your example,  you want to find hex values.  I'd rather see a lognorm
>>> "parser" created for this purpose.   I'm afraid if we start with regular
>>> expressions,  we'll end up mixing rules (RE and non-RE) and it will make
>>> things very confusing.  I really like the way lognorm's "masking" works.
>>> I'd rather not see that break.
>>>
>>>
>>>
>>> On 11/03/2014 07:15 AM, singh.janmejay wrote:
>>>>
>>>> We log a warning?
>>>>
>>>> --
>>>> Regards,
>>>> Janmejay
>>>>
>>>> PS: Please blame the typos in this mail on my phone's uncivilized soft
>>> keyboard sporting it's not-so-smart-assist technology.
>>>>
>>>>
>>>> On Nov 3, 2014 5:37 PM, "David Lang" <david at lang.hm
>>> <mailto:david at lang.hm> <david at lang.hm>> wrote:
>>>>
>>>>     On Mon, 3 Nov 2014, singh.janmejay wrote:
>>>>
>>>>         I am thinking of it as a 2nd class field-type.
>>>>
>>>>         By that I mean, one gets best performance from 1st class
>>> supported
>>>>         field-types, but if for some reason that is not sufficient for
>>> someone,
>>>>         they can use a regex-field-type. It may be a little low on
>>> performance, but
>>>>         then it unblocks people immediately.
>>>>
>>>>         I can do it, just need to know we are not ideologically against
>>> it.
>>>>
>>>>         Kind of construct im thinking of:
>>>>
>>>>         %foo:regex:[a-f0-9]+% to match hex-numbers for instance.
>>>>
>>>>         Thoughts?
>>>>
>>>>
>>>>     Since this will be such a performance pig compared to the existing
>>> parse tree, how about requiring a 'enable low performance types' flag or
>>> something like that to enable it?
>>>>
>>>>     There needs to be some good indicator that this is a performance
>>> problem.
>>>>
>>>>     David Lang
>>>>     _______________________________________________
>>>>     Lognorm mailing list
>>>>     Lognorm at lists.adiscon.com <mailto:Lognorm at lists.adiscon.com>
>>> <Lognorm at lists.adiscon.com>
>>>>     http://lists.adiscon.net/mailman/listinfo/lognorm
>>>>
>>>>     _______________________________________________
>>>>     Lognorm mailing list
>>>>     Lognorm at lists.adiscon.com <mailto:Lognorm at lists.adiscon.com>
>>> <Lognorm at lists.adiscon.com>
>>>>     http://lists.adiscon.net/mailman/listinfo/lognorm
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> Lognorm mailing list
>>>> Lognorm at lists.adiscon.com
>>>> http://lists.adiscon.net/mailman/listinfo/lognorm
>>>
>>>
>>> - --
>>> - - Quadrant Information Security
>>>   Champ Clark III
>>>   o: 800.538.9357 x 101
>>>   c: 850.443.2440
>>> -----BEGIN PGP SIGNATURE-----
>>> Version: GnuPG v1.4.11 (GNU/Linux)
>>>
>>> iQEcBAEBAgAGBQJUV6S0AAoJENnmXt7Lmc3KAwEH/jiwhj/nhFRpvRm7DttrvYTE
>>> U7kHpjToIMlCSiJqiS4bnvOTS4oTG6mQ5myYzAr16ITTIJQLnJSHmVWBgxlGdUlM
>>> Kq33I+zYjfUK2Go01PSoLjoE3rgdGa1hptFHVUZREuRkzSP6THtXn8XhexmZzdpH
>>> 1lC+ZXwNZ9k7FsWm3M027I8zGDKnvZLVdf2UJUElyrNxmWW04ieR3lqJ5qh3Uj8l
>>> OvTzEDYObghwyS6hThNb1oMz2Sr1AD/mWu+sri1BDKqlPyMkQgYe8KNyI4sKBQHL
>>> AdCi3TXMvYa8WMb5AOVL4tX0QrOgwNj5mbpeWb2vWrV/S2mkN39TZbE406W91T8=
>>> =rGsk
>>> -----END PGP SIGNATURE-----
>>>
>>>
>>> _______________________________________________
>>> Lognorm mailing list
>>> Lognorm at lists.adiscon.com
>>> http://lists.adiscon.net/mailman/listinfo/lognorm
>>>
>>>
>
-------------- next part --------------
_______________________________________________
Lognorm mailing list
Lognorm at lists.adiscon.com
http://lists.adiscon.net/mailman/listinfo/lognorm