[Kea-users] 1.3.0 beta not working with mysql

[Bill Pye]

Hi Thomas

Thanks for your follow-up. While I said that my installs of Percona & MariaDB were standard I did forget that I'd changed teh character set & collation to "urf8" & "utf8_general_ci" respectively. Just in case this had any effect (I know it shouldn't but...) I decided to install a new MariaDB totally basic with nothing else but the kea DB in it - this was on one of my current servers. I've spent the week-end trying to get something that made sense but not much luck, I'm afraid.

Today I did some more tests on this single MariaDB server and basically came to the conclusion that the only version of kea that works using the db is kea-1.0. My last test a short while ago was with kea 1.3 built with the srpm provided by Rasmus Edgar, that didn't work and seemed to give the same error that I've mentioned before. I have a debug log file if that's of any use to you?

Is there anything specific that I can do to get more information about this problem for you? Meanwhile, I could try with a more basic clean install of CentOS7/MariaDB and see if I get the same results as you.


Regards



Bill

----- Original Message -----
> From: "Thomas Markwalder" <tmark at isc.org>
> To: "kea-users" <kea-users at lists.isc.org>
> Sent: Monday, 9 October, 2017 14:29:37
> Subject: Re: [Kea-users] 1.3.0 beta not working with mysql

> Hi Bill:
> 
> Just following up with you.  I was able to build out a Centos 7 VM with
> MariaDB.  Kea builds fine, the unit tests all pass (which include pretty
> exhaustive testing of our supported backends).   kea-dhcp4 comes and
> happily hands out leases with lifetimes of 604800 with no apparent
> issues.   This was using the 1.3-beta code as is.   I even verified that
> setting the wait_timeout to 30 * 86400 seems to work fine using the
> mysql command line client.
> 
> I can't say why you saw what you did before.  Doesn't make much sense.
> If you see it again let us know and we'll dig further.
> 
> Regards,
> 
> Thomas Markwalder
> ISC Software Engineering
> 
> On 10/6/17 1:33 PM, Thomas Markwalder wrote:
>> It's stored as an unsigned integer, so in theory it's max is
>> |4294967295.  Certainly much larger than 604800.  For grins, I used that
>> value on a machine running MySQL and it inserts leases just fine.
>>
>> Using the unsigned int max value, though  blows up time calculations
>> which manifests itself as this error:
>>
>> 2017-10-06 13:29:57.202 ERROR [kea-dhcp4.alloc-engine/1888]
>> ALLOC_ENGINE_V4_ALLOC_ERROR [hwtype=1 08:00:27:99:72:be], cid=[no info],
>> tid=0x489a9: error during attempt to allocate an IPv4 address: Could not
>> create bind array from Lease4: 178.16.1.100, reason: Time value is too
>> large: 5802278292
>>
>> Clearly different from your whose reason is (error code 0).
>>
>>
>> |On 10/6/17 12:37 PM, Bill Pye wrote:||
>>> Hi Thomas||
>>>
>>> An update on my 'problem' but let me start with a question - is there any
>>> maximum value for the "valid-lifetime" setting? I ask this because I've just
>>> noticed a slightly odd number set for that option and it was set as "604800", I
>>> can't imagine why I set it that high or if it was a typing error. I've set that
>>> to a more reasonable value of "86400" and restarted my kea 1.3 servers and
>>> pointed them to my Percona cluster, lo and behold they came up with no errors
>>> (all three of them). They're all sitting there responding to DHCP requests, to
>>> say I'm surprised is an understatement!
>>>
>>> I'll leave these running overnight and see what happens, is it worth upping the
>>> loglevel a bit? I currently have logging at "INFO" and a debug level of zero,
>>> which is the best for reasonably detailed info but doesn't flood the server?
>>>
>>>
>>> Regards
>>>
>>>
>>> Bill
>>>
>>> ----- Original Message -----
>>>> From: "Thomas Markwalder" <tmark at isc.org>
>>>> To: "kea-users" <kea-users at lists.isc.org>
>>>> Sent: Friday, 6 October, 2017 17:16:15
>>>> Subject: Re: [Kea-users] 1.3.0 beta not working with mysql
>>>> Hi Bill:
>>>>
>>>> Ok, thanks.  I'm building out a Centos 7 VM with MariaDB.   I personally
>>>> have have tested 1.3 agains MySQL and Postgres.  Not MariaDB  though.  I
>>>> know the latter is "supposed to be binary replacement for MySQL" ... but
>>>> sometimes reality is less than ideal ;).
>>>>
>>>> Can you tell me if there is any substantial amount of time elapsing
>>>> between server startup and the first client attempt?
>>>>
>>>> Thanks
>>>>
>>>> Thomas
>>>>
>>>>
>>>> On 10/6/17 10:57 AM, Bill Pye wrote:
>>> _______________________________________________
>>> Kea-users mailing list
>>> Kea-users at lists.isc.org
>>> https://lists.isc.org/mailman/listinfo/kea-users
>>
>> _______________________________________________
>> Kea-users mailing list
>> Kea-users at lists.isc.org
>> https://lists.isc.org/mailman/listinfo/kea-users
> 
> 
> _______________________________________________
> Kea-users mailing list
> Kea-users at lists.isc.org
> https://lists.isc.org/mailman/listinfo/kea-users