Lease time (F.A.O. Chris Pettitt)

[#1]

Quote from: Anotherone

it might get re-allocated to another user?

No.

[#2]

Quote from: Chris

even if the previous connection only lasted a few seconds.

Yes, that's what I thought.

[#3]

So Keeef ("can I stay connected for longer than 60 hours?") we know that an IP address gets changed with every new PPP session, no matter how short, and my claimed record for an unbroken PPP session is around 900 hours (on ADSL2+ if that makes any difference).  Have we answered the question or are we still playing 'guess what the problem is'?
You might also be interested in this thread http://community.plus.net/forum/index.php/topic,108955.16.html where it appears that regular daily disconnections may have been caused by a router setting (see reply #18).

[#4]

Quote from: ReedRichards

Have we answered the question

Not yet.

[#5]

So you are looking for an exact figure?
I don't know if this information (from http://community.plus.net/forum/index.php?topic=35722.15;wap2 ) is correct but the assertion is that there is no lease time assigned.

Quote

benoh: Hi,     FYI, IPs given out on dial/dsl either static or dynamic are assigned via PPP and not DHCP, there is no "lease time", if the IP is dynamic then the RADIUS (Remote Authentication Dial In User Server) will take the first free IP from the configured pool and assign it to the end user. It could be made to have a lease time but there isnt much point, when the PPP session terminates the IP is free and hence can be used by anyone trying to connect. Dynamic IPs on plusnet (dial at least) do not have rDNS associated to the username currnetly using it, although this could easily be achieved with dynamic DNS updates directly from RADIUS without too much overhead.

[#6]

Quote from: ReedRichards

the assertion is that there is no lease time assigned.

If that is correct. Would it not be a good idea to have a short one?
100 seconds should give a router time to log back in should a power spike cause a reboot for example.

[#7]

No, it wouldn't be a good idea, it would be an unnecessary overhead which involves processing power and time and that costs. I'm not bothered by Dynamics IP's, nor am I prepared to pay more for my broadband just to satisfy this sort of thing. If you want a static IP you have the option to pay for it, as I would if I wanted or needed a static IP.
It is already this processing power and time (sampling rates etc) that occasionally mean that a drop in PPP session isn't picked up by the RADIUS logs. There have also been incidents where an IP address hasn't changed after a PPP drop, but both of these events are not common.
They have been explained by Plusnet in the past as "because the duration of the drop has been short" so there may be in effect be some sort of lease time but purely because of the way the system actually runs.. (Some modem/routers - usually the more expensive ones with more processing power and better firmware have the ability to negotiate connections more quickly than the average, and some are slower.)
To go back to the points raised earlier by RR, you need to be very careful about distinguishing between 1) System Uptime (ie. the time since the last power cycle/reset) and 2) Line Uptime (the time you are in sync) which may be the total time in sync, often a Line count is given which is the number of sync events - WAN uptime can be exactly. the same as can Showtime, and 3) On Line Time / Total On Line Time which can indicate the total length of all the PPP Internet sessions time since the last reset/power cycle/reboot.
A lot of exactly what level of information is available on/for a given modem/router will depend on the individual firmware and varies from manufacturer to manufacturer.
So I doubt very much whether 900 hours was actually the duration of a PPP session, in today's "environment" with the "Network" maintenance and failures that have occurred in recent times both within the BT infrastructure and within Plusnet, you have had to have been extremely lucky not to have at least one PPP session drop which you may not even be aware of.
I know when I've had a PPP drop, because I've set my modem/router not to connect automatically, but it doesn't tell me a PPP session count only the total PPP time. However I've just had in excess of 980 hours of Line Uptime continuous sync, which would have been 4 days longer had I not forced a resync but I know there has been at least one PPP drop and I may have Gateway hopped a while ago - I don't keep detailed notes unless I've got an issue, just periodic RS SNRM graphs and DMT stats grabs.

[#8]

Quote from: Anotherone

No, it wouldn't be a good idea,

I wonder how many/what percentage of calls/questions are because of disconnects?

[#9]

Quote from: Anotherone

I doubt very much whether 900 hours was actually the duration of a PPP session, in today's "environment" ...

Good point; here is all the information I have:
I have a Netgear DG834PN router and the log is set to record all events but this data is lost if the power fails.  So my current log goes back to 1st November and shows three subsequent events, each culminating in a successful CHAP authentication, since that time.  The WAN uptime is shown as 56 hours which is the time since the last such event.  I assumed that CHAP authentication was part of the necessary process in re-establishing a PPP session and therefore what is reported as WAN uptime is the length of the current PPP session; is that correct?
The power cut was on Wednesday 31st October.  A few days prior to that, I don't recall exactly how many, I noted that the recorded WAN uptime had passed 850 hours.  I had been watching this since my target SNR was set to 3dB and was amazed at how stable the connection had appeared to be. 

[#10]

Quote from: ReedRichards

I noted that the recorded WAN uptime had passed 850 hours.

Truely amazing!
I'd be happy with that.

[#11]

As I mentioned earlier, that could be total ie cumlative WAN time, not necessarily a single connection. Yes CHAP authentication is part of re-establishing a PPP session.
Whilst I obviously know of that Netgear model RR, I don't know it from direct experience, but I suspect it is total time. Getting a resync without rebooting or power cycling would be the way to find out if you really want to know, but as you may not wish to do that for obvious reasons make a note of your sync speed and keep a careful watch. If/when you see a change in sync speed, check if the WAN time has been reset. Likewise, do a Gateway hop (ie drop your PPP session and reconnect) I doubt the WAN time will reset. That is the only way to be sure, as sometimes these things can vary from model to model even with the same manufacturer. (Similar comments can be made about error figures and when they get reset).
Keeef, the number of calls/tickets about disconnects will be high and is usually because of a loss of sync (not always the case), but whether the IP address has changed is totally irrelevant (as would be any "lease time" if it existed) to the cause of the drop and what was needed to fix it if it was a fault.
Edit:spelling

[#12]

Quote from: Anotherone

the number of calls/tickets about disconnects will be high

So it would make sense to try and reduce these.

[#13]

I'd guess at lease 95% of Plusnet's users wouldn't have a clue what their IP was and even if they did know couldn't care less if it changed or not. Of those that do know a lot of them can get along using something like DynDNS.

[#14]

Quote from: Keeef

So it would make sense to try and reduce these.

If you have some clever way of reducing the number of overhead and underground cable faults I'm sure OpenRetch will offer you a job.
These will be a high proportion of the ticket/calls.

[#15]

Quote from: jelv

I'd guess at lease 95% of Plusnet's users wouldn't have a clue what their IP was

LOL!