Switches, Hubs, and Modems
cancel
Showing results for 
Search instead for 
Did you mean: 

experience strange behavior

Vanessa_14
Advisor

experience strange behavior

Hi I'm experience strange behavior.

Previous my subnetmask was 255.255.255.0.
I extended the subnetmask to 255.255.0.0.
Because I need more IP adresses.

My servers and Clients are in the 192.168.0.x range. My Ilo's I configured in the 192.168.1.x range contected on a seperate Switch that is uplinked to the other server swithes.

I can successfuly ping from a test PC with IP 192.168.1.0 to my ILO's configured with an adres in that range. ALways!

From the servers I somethimes looses connection pinging to 192.168.1.x range.
Resetting the swith where the ILO's are connected to solves the problem again, temporalilly. But after a while the looses again connection...

My server are as stated before configured in the 192.168.0.x range / 255.255.0.0.

I've changed the swithes and experienced the same behaviou where the ILO's are connected so I don't think my swithes are bad.

Al my swithes are Intelligent Managable Cisco Swithes. Could this be a temporary problem, because some swithes still "think" my subnetmask is 255.255.255.0 instead of 255.255.0.0?
11 REPLIES
Shiraz Malik
Advisor

Re: experience strange behavior

You have all "Intelligent Managable Cisco Switches" ???
Problem is really interesting to resolve, but i'm confused if you have all cisco switches, then what makes u come here in HP forum?
Vanessa_14
Advisor

Re: experience strange behavior

Hi,

Because I experience the problem with ILO's.
All my server are HP and Compaq Servers.
Another reason is because cisco hasn't such a great forum.

Not al my swithes are Cisco's but most, are...

Thank you.
OLARU Dan
Trusted Contributor

Re: experience strange behavior

Vanessa,

I think it is definitely a routing problem, not a switching problem. Has this ping problem occured before you changing the subnet mask?
Vanessa_14
Advisor

Re: experience strange behavior

Hi, dan, thanks for the reply.

No before changing the subnetmask I never experienced strange network behavior.

But the weird thing is.

Previous my network range was 192.168.254.x with 255.255.255.0.

I changed everywhere i could think the subnetmask to 255.255.0.0.
(Routers, servers, DHCP ranges, IP's from the swithes, printers etc etc...)

Now I started to use 192.168.0.X adresses.
3 new ILO's are added and one test PC
192.168.0.1 till 192.168.0.4.

I can always ping from the test PC to all my ILO's.

I start the same ping from a server 192.168.254.1 to my 3 ILO's 192.168.0.1 till 192.168.0.3 and after a while 2-5 minutes the pings stops responding from this server, but still works further on my test Pc with ip adres 192.168.0.4.

I'm in the possibility to ping from 192.168.0.4 to my three ILO's. But after a while I can't ping from 192.168.254.1 to my 3 ILO's, rebooting the swithes resolves the issue for another 5 minutes.

Strange Isn't It? Could this be a problem with a swith learning its ipadresses, and became a little bit confused about the subnetmask change, I know they keep some "ip- table list" that its automaticly filled up by learning IPadresses?


Any Idea?
Les Ligetfalvy
Esteemed Contributor

Re: experience strange behavior

I don't claim to be an expert in subnetting and supernetting, but the notion of taking what is basically class 'C' 192.168.y.z scopes and joining them together does not sit well with me. I suspect that some device somewhere is still respecting the 'C' mask and may be hard-coded to do so.

What do liberal amounts of "route print" reveal?

Why not stick with routing between the subnets or change to a 172.x.y.z address scheme?
Vanessa_14
Advisor

Re: experience strange behavior

It is almost impossible to change to 172.16.x because some servers running specific software where the licence is tied to the Ip adress...

Routing between them makes that router to be single point of faillure in case of a disaster.

OLARU Dan
Trusted Contributor

Re: experience strange behavior

Have you tied "ip subnet-zero" and "ip classless" on your router?
Tv_3
Respected Contributor

Re: experience strange behavior

Hi,

You are violating the basic network concepts!!!

192.168.x.x is a subnet rage CLASS C with means that you have 255 segments of 255 host, meaning that each segment cannot count more that 255 host. You need to couple segments with routers, routing from one segment to another.

I think you'd best leave the network applicances that cannot be changed from IP adres to leave them where they are. Add other network appliances to a totaly new Ip range (172.16.x.x/16) or (10.x.x.x /8).

Put a router between the two segments.

For performance you can consider setting up two routers. ARouter from A to B and a router from B to A.


Chaos, Panic or Disaster??? Always Stay Calm, I will fix it.
Antoniov.
Honored Contributor

Re: experience strange behavior

Hi Vanessa,
I don't understand if you changed mask on all hosts and switches of your network. Hosts without changes can't work in network. Don't forget you are using non standard subnet mask, so if you have some old device it can't work.

Antonio Vigliotti
Antonio Maria Vigliotti
André Beck
Honored Contributor

Re: experience strange behavior

Tv wrote:

> You are violating the basic network
> concepts!!!

By what, applying a prefix shorter than /24 in an ex-Class-C range?

> 192.168.x.x is a subnet rage CLASS C

Class C (together with all other classful adressing) has been abandoned more than 10 years ago and can be considered properly buried today. And when we are at it, strict RFC950 subnetting was outdated for nearly as long as it exists, 18 years IIRC. If he wants to use 192.168/16, nothing except extremely old and never updated hard-/software might choke on it. And this is more a reason to replace this hardware than it would be to reintroduce late concepts which have been kicked deliberately.

> with means that you have 255 segments of

256

> 255 host, meaning that each segment cannot

254. Unless you are really up to date with IETF matters and consider setting free the all-zeros host part an idea you must try out in a production network instantly ;)

> count more that 255 host. You need to
> couple segments with routers, routing from
> one segment to another.

You can do this. But you don't need to if you can migrate everything cleanly to 192.168/n (16 <= n < 24) , which means you either can change netmasks everywhere or at least everywhere it counts (one can collect unchangeable old and broken end systems together with all the systems that need to talk to them into a /24 subrange so the rotten stuff can ARP for the relevant addresses).

What Vanessa describes (losing connectivity after approx. 5mins) actually sounds like a bridge table issue, either some MAC bouncing between VLANs or starving unicast traffic together with a bug that prevents it from flooding orderly. Switches are not involved into IP routing at all, they are bridges - unless we are speaking of L3-Switches, which are routers. Anyway, if it is an ARP issue, Cisco would make it a problem occuring after approx. four hours ;)

BTW, any DHCP active in that broadcast domain?

Andre.

Re: experience strange behavior

I completely agree with Andre but like to add that IP routing in that case still makes sense to limit the broadcast domain.

using the cisco switches provides presumably thebasic routing functionality to support a couple of static routes for different 192.168.x .y networks.

classless subnetting and changing from one range to a bigger one usually busts since I tend to forget some network masks or some routes on the not "central" devices.

the router/single point of failiure does not count since there are many different options to make routing switches highly available. from VRRP and flavours over some ospf constructs with differen transport networks. whatever.

regards.

frank