[kea-dev] Lease File Compression Requirements
marcin at isc.org
Thu Dec 18 15:22:31 UTC 2014
See my answers inline.
On 12/18/14 11:25, Chaigneau, Nicolas wrote:
> I've read the latest version (3) of the requirements page (and the related messages posted on the list).
> This seems good to me.
> A few points I'd like to discuss:
> On the two vs one lease file debate:
> For me, having two lease files instead of one really is not an issue. Administration will adapt.
> Performances, however, are crucial. So if there's a noticeable gain to expect from the multiple files option, I'd prefer to go that way.
I agree. I will update the requirements with the decision that the
server doesn't have to keep leases in one file. And, the design will
> On requirement 2.b:
> "The LFC must be triggered periodically and the interval must be configurable."
> One additional thing that could be handy is to have LFC occur at predictable times (independently of when the server was started).
> For example, the ability to trigger LFC once an hour at the beginning of each hour.
> Maybe something similar to cron scheduling ?
Perhaps the server should persist an information about the time when the
last LFC occurred and measure the period from there. And if the server
failed to write the last LFC time, e.g. because of the crash, the LFC
could start immediately or nearly immediately.
> Following point 2) above, it could be useful for an external process to know when LFC is currently happening.
> (so it would know, for instance, that it is not advisable to read the lease file until LFC processing is finished)
> This could be achieved for example through a "lease lock file", created and deleted by Kea, respectively at LFC start and completion.
Yes, I was thinking about it before. I think it is necessary for the
server to know whether the last cleanup completed successfully or not
when the server crashes in between. So, I agree.
> On lease file synchronization across servers (in active/standby mode):
> This can be taken care of outside of Kea.
> For instance, this can be achieved through a HA cluster architecture with DRBD.
> On 16/12/14 12:02, Marcin Siodelski wrote:
>> As a first step towards the implementation of this feature in Kea, I
>> created a page with requirements:
>> All input to this document is welcome.
>> I am planning on creating an accompanying design document using the
>> requirements presented here.
>> Again, please share your thoughts.
> This message contains information that may be privileged or confidential and is the property of the Capgemini Group. It is intended only for the person to whom it is addressed. If you are not the intended recipient, you are not authorized to read, print, retain, copy, disseminate, distribute, or use this message or any part thereof. If you receive this message in error, please notify the sender immediately and delete all copies of this message.
More information about the kea-dev