I having a problem with a small network with seven workstations and two network printer/scanners. The scanners have Scan-to-PC capability, with an address book of each user by IP address. Occasionally, all the IP addresses on the network will change as if they have been reassigned, so Scan-to-Pc either doesn’t find the destination user, or it sends it to the wrong user, because the IP address has been reassigned.
The first couple of times this happened, I concluded that this was occurring because of an unnoticed overnight power outage, and when the power came back on, the router was reassigning IP addresses to the workstations in the order they powered on, causing the change from the assignments before the outage.
I put the router on a UPS and this seemed to solve the problem for a few months, but it just happened over the weekend again. I asked the users if they had noticed any clocks that needed resetting (coffee maker, microwave), and the only thing we noticed was a microwave oven on a GFCI socket that had tripped. I am thinking there was another power outage, but this time, it exceeded the back-up time provided by the UPS, and the router lost power.
Does this sound likely? If so , I will just buy a larger UPS that will extend the back-up time to reduce the likelihood of power loss to the router. But I don’t want to buy new hardware if I am making incorrect assumptions that may not fix the problem.
Rather than go nuts trying to preserve every last second of power for your router, it would be a lot easier to simply assign static IP addresses to the workstations, or configure your router to hand out a preconfigured IP for each MAC address.
Some assumptions I’m making here that may be incorrect: I assume your ‘router’ is assigning dynamic addressing via DHCP. I assume you have the default address lease duration that shipped with the ‘router’.
Ok, assuming that’s correct what is happening is that when the power goes out the devices can’t update their lease with the ‘router’ and so request and are granted a new lease which is assigning them a new address when you lose the power.
You could fix this either by increasing the lease duration (IIRC, most of them have like an 8 hour lease as the default) to something like 30 days…or you could simply assign your devices static IP’s. I would probably go with the latter if IP address is critical.
Hope this helps and I understood your question…sorry, doing like 5 things at once here.
Going to static IP addresses makes sense, but I am not sure if the network printer/scanners support it. I will have to do some research on that, though if the user addresses are static, the printer can probably find them.
Most dhcp servers (even those included on soho routers) can do ‘static’ dhcp. There’s usually a section where you can enter in the mac addresses and address reservations for your equipment so that the server hands out the same ip address every time. That way if DNS entries or the gateway or whatever ever changes, you can just make 1 change instead of changing all the computers on the network. I wouldn’t mess with static ip addresses unless it’s absolutely necessary. Any kind of company growth and it’ll turn into a management nightmare.
You should be able to…most print servers allow you to configure them for static as well as DHCP. It probably has a web interface (should be in your documentation or you can look it up online), so just use the current IP and your browser to access it and check the Static IP box…then just give it an IP outside your DHCP range (assuming you still want to use DHCP at all).
Or access your router and change your lease duration…that’s the other option. Again, should be easy enough to do…most use web access for administration these days.
Good luck! If you have any other questions, don’t hesitate to ask.
Extending the lease would be my first suggestion, but I did something very similar long ago on a server-based network and what I did was have the scanner upload the images to a server - each user had a private area - and the server ran an OCR package which produced text on appropriate documents - you simply hit the right buttons on the scanner. Perhaps that might be an alternative?
I am looking at my Linksys router Setup/Basic Setup page, and the default setting for Client Lease Time is “0” minutes, which defaults to one day. Shouldn’t that be enough to handle a short outage of 30 minutes?
You should be able to assign static IP addresses to the workstations and if it is to much of a hassle to get the scanners and printers to have static addresses let DHCP give them IP addresses. I do this at home. The Wii, laptops and the Roku music box get dynamic addresses and the desktop and NAS get static addresses. It sounds to me the problem you have is the scanners look things up based on IP address and not names.
I’m wondering if long lease times will help, because if the clients rebooted during the power outage, or were shut down overnight, and the DHCP server lost its database during the outage, the clients will not receive a response when they try to renew their lease, which might prompt them to initiate an entirely new DHCP request (i.e. not requesting the same IP address).
In which case the only solutions are static addresses or a bigger UPS.
Should, but the way it works is the client re-negotiates with the DHCP server after some fixed time (that is less than the full lease). So, if you had a power outage at the start of the day you would probably be fine…but if you had one around the time the client would need to re-negotiate then it would ask for a new address on reboot.
Though I would think it would just get the same address it had before to be honest. Still, you could try either the suggestion of using DDNS with dynamically assigned static addresses or you could try either direct statics or simply making your lease for several days (I usually do 8…it cuts down on traffic between client and server).
With longer leases, and even with a bigger UPS, there will be situations when IP address assignments can change. There is a fundamental conflict between dynamic address assignment from an address pool and the scanner’s use of fixed IP addresses to identify particular hosts or users. Whether they’re meted out by a DHCP server or manually configured, assigned IP addresses would seem to be the only way to reliably solve this problem. IMHO, a bigger UPS just moves you a from the current state of “mostly works” to “mostly works even more.”
Yes, this is really the best solution for the reasons mentioned. I notice people are still suggesting assigning static IP addresses to the PCs or extending the lease. spazattak’s suggestion is much, much better.
I’m not seeing the problem people seem to have with statically assigning IP addresses to devices. To be sure, if you want to screw around with MAC or port based IP assignment (assuming you have the capability for it), knock yourself out. Why this is a supposed ‘better’ solution is beyond me though, unless you having planned out your IP address architecture well and feel you are going to junk it and re-do the entire thing in the future.
Even in very large IP networks I’ve designed I use static IP addresses all the time for network devices. Since we are most likely talking about RFC 1918 addressing here it’s not like we are paying for address space on a LAN or anything…sheesh. As for administration, it only comes into play if either A) You don’t actually have any idea of how your addressing is allocated, or B) You plan to radically change your logical address architecture some time in the future.
My logical IP address architecture specifically designates IP addresses for the express purpose of being static in fact. Sure, if the customer has the capability for VLAN or MAC based IP assignment I would use it…but if not, I wouldn’t. For a small little network and a couple of scanner/printers where it’s unlikely to grow beyond a class C? I wouldn’t bother, personally. It’s more work than it’s worth to simply input the MAC addresses IMHO. YMMV of course…but it’s not like it’s some kind of crazy idea to use statics, people.
As for using dynamic addressing and changing the leasing, it’s what Fear already had…the suggestion was to ‘fix’ the problem he originally was asking about in the OP.
I took a quick look at the manual for the KM-3035. Take a look at page 3-63; it indicates that you can enter a host name rather than an IP address. If so, another option is to take gazpacho’s suggestion, and arrange for DHCP-assigned addresses to be registered with your DNS server. Depending on how fancy you want to get, and how convoluted a solution you want, you could even use this capability to arrange for your users’ scans to follow them to any PC they chose to log in to.
In case it wasn’t clear, I was doubting that longer leases would make any difference, because the Linksys box would presumably have lost its DHCP data in the power outage.
Actually, it seems that it can be any name you choose to enter in the scanner’s Address Book with the Address Editor utility. Normally, it would be the computer name. According to the manual, either DNS or WINS can be used to resolve names.
The normal sequence of events would be something like this:
[ol]
[li]Use the Address Editor to create a destination associated with a host (computer) name.[/li][li]Host boots, or for some other reason, acquires new IP address from DHCP server[/li][li]Host registers host name with DNS or WINS server. (I know next to nothing about WINS, but I assume dynamic registration is possible.)[/li][li]User scans something choosing the destination from step 1.[/li][li]Scanner looks up the host name associated with the destination name in its address book.[/li][li]Scanner looks up IP address for host name using DNS or WINS, returning IP address registered in step 3.[/li][li]Scanner sends resulting scan to that IP address.[/li][/ol]
A more sophisticated, but perhaps unnecessarily complicated, approach would be to create a subdomain like users.mycompany.com, and arrange for some sort of login script to register a name of the form username.users.mycompany.com upon login. Then you could set up a destination in the scanner for each user that would direct their scans to the machine on which they had most recently logged-in.
Missed the edit window, and wanted to add: If you don’t require the scans to be sent to individual computers, and have a central file server that can serve as a repository for them, I think Quartz’ suggestion is a good one. Having them sent to access-controlled directories on a single host, and arranging for users to have access to those directories eliminates a lot of complexity.
As for administration, it only comes into play if either A) You don’t actually have any idea of how your addressing is allocated, or B) You plan to radically change your logical address architecture some time in the future.