FAQs

1. The PC is directly-connected to a switch
2. The PC is running Novell's VLMs or Client32
3. The switch has Spanning Tree Protocol (STP) enabled

In this situation, when the directly-connected PC is powered on, the switch senses linkbeat on that port. This causes the switch to go through the four Spanning Tree states: blocking, listening, learning, and forwarding. It takes 30 seconds for the switch to complete that sequence and begin forwarding packets to and from that port. During those 30 seconds, Novell sends 3 requests for a server, then stops looking. By the time Spanning Tree completes its job, Novell reports that "a file server could not be found".

There are several workarounds available:

1. Disable Spanning Tree on the switch (if Spanning Tree is not needed, i.e. no loops in the network topology).
2. For VLMs, add a "pause" just after calling VLM.EXE in STARTNET.BAT. When the user reboots a PC, have them wait at least 30 seconds before continuing the sequence. This workaround is documented on Novell's Knowledgebase (www.support.novell.com, search for document 2920460).
3. For Client32, add a registry entry in the PC, as documented on Novell's Knowledgebase (search for document 2925582).

In the Fall of 1998, HP is planning to release switch firmware with an enhancement to resolve this timing problem between Novell and STP. The enhancement allows users to configure Spanning Tree so that it does not go through the 4 states, on a port-by-port basis. Instead, for those configured ports, Spanning Tree will immediately begin forwarding packets to and from the port. This allows Novell clients to communicate with the server as soon as the network card (NIC) is enabled. After that, the switch continues to listen for and send Spanning Tree packets on those configured ports. This protects the user who might inadvertently connect a hub or switch to that port and create a network loop—Spanning Tree will detect the loop after a short time, since the port listens for and sends STP packets on that port.

» Return to top

Q: What is port trunking, and how do I configure it?
Port trunking is connecting two switches with multiple links that act as one high-bandwidth link. Up to four ports (full- or half-duplex) can be used for each trunk on a Switch 2000 or Switch 800t. This enables up to 800 Mbps of bandwidth for linking switches. The benefits of port trunking are:

• Increased bandwidth between two switches
• Redundant links between two switches

The illustration below shows how port trunking is used in three different locations. Initially, all of the links were 100 Mbps links at 100TX speed and in full duplex mode. Notice how much the bandwidth increases by setting up port trunking between the switches.

Figure 1

How to Configure the Switches for Port Trunking

To increase your bandwidth between two switches, select the ports on two switches in your network where you want to implement port trunking. When you select the ports, all of these ports in a trunk must be of the same link type (such as 10Base-T or 100vg) and all should use the same mode (full- or half-duplex).

To increase bandwidth between your switches, follow these steps:

1. Connect your workstation to the VT-100 console on a Switch 800t or Switch 2000 using Telnet, a serial cable connected to a console or HP Advancestack Assistant.
2. From the Main menu, select Configuration.
3. In the Configuration screen, select Ports.
4. Assign a port to the trunk by following these steps:
        1.   Select Edit.
        2.   Move the cursor to the Trunk column.
        3.   Select the row for a port you want in the trunk.
              ress the spacebar once to select Trk1 for trunk 1.

Example. If you used these steps to assign ports A1, A2, and A3, here's how the screen would appear after they have been assigned to the trunk Trk1:

• If you plan to move a trunk port that is used in conjunction with filters and port monitoring, the filters and port monitoring must be reconfigured accordingly.
• After you configure a port trunk, these ports appear as a single port (labeled "Trk1") on other configuration screens, such as the Spanning Tree and Port VLAN assignment screens.
• If you have switches connecting two buildings, we recommend that you pull the cable through different paths between the switches. If one link in a port trunk fails, the other ports in the same port trunk will route the traffic successfully.

» Return to top

Q: How do I connect the HP 100Base-T Hub-12tx/TXM to an HP Switch 800t?
The Advancestack Hub-12tx/TXM can be connected to an external Switch 800t. The maximum distance is 100 meters from the hub to the switch using the MDI port. The hub and switch connection improves performance because of two reasons:

• Adding switches to an existing network increases network diameter.
• Each hub is on its own collision domain and is no longer carrying traffic for all of the other hubs connected to it.

To improve your network, follow these steps:

1. Connect port number 12mDI from each Hub to any port on the Switch 800t using RJ-45 TP Category 5 cable up to 100 m.
2. For each hub connected to the Switch 800t, check that the "Link/Traffic" LED is lit on port 12.
3. To prove that the connection between the switch and hub is working, retrieve and open a file from another server connected through the Switch 800t. For example, in the above illustration a manufacturing user would copy a file from the Engineering Server. For example, in the above illustration a manufacturing user would copy a file from the Engineering Server because the Switch 800t is between the user and the server.
4. Using the VT-100 console connection, from the main menu of the Hub select Statistics, then Port Statistics.
5. Verify that port 12 is operating correctly by checking to see if any errors occurred during transmission of files. If errors do occur and you have followed the above steps correctly, you may just have a bad cable. Try another cable.

» Return to top

Q: How do I connect the HP 100Base-T Hub-12tx/TXM Switch Port Module to an HP Switch 800t?
The Advancestack Hub-12 TX/TXM Switch Port Module (TX or FX) can be connected to an external Switch 800t. The maximum distance for TX is 100 meters from the Hub to the Switch and 2 kilometers for FX with both ends set at full duplex. The Switch Port Module allows your network to increase the bandwidth to full duplex and connect groups between buildings with 2-kilometer fiber cable.

To connect ports on the Switch Port Module to the Switch 800t, follow these steps:

1. In the Hub-12tx or Hub-12txm, there is Switch Port Module with a TX or FX port. Connect this port to the Switch 800t using Category 5 UTP/STP cable or fiber-optic cable, respectively. Check that the "Link/Traffic" and "FDX" LEDs are lit.
2. To prove that the connection between the switch and hub is working, retrieve and open a file from another server connected through the Switch 800t. For example, in the above illustration a manufacturing user would copy a file from the Engineering Server because the Switch 800t is between the user and the server.
3. Using the VT-100 console connection, from the main menu of the Hub select Statistics, then Port Statistics.
4. Verify that the "Exp Port" is operating correctly by checking to see if any errors occurred during transmission of files. If errors do occur and you have followed the above steps correctly, you may just have a bad cable. Try another cable.

» Return to top

Q: If the MAC address table overflows by receiving more addresses than the switch address table can hold, how will the switch treat incoming packets whose addresses are not in table (addresses which have not been learned)?
The switch floods (broadcasts) the packet to all ports, the same as a bridge would do.

» Return to top

Q: Can the Switch 800t's STP (Spanning Tree Protocol) parameters be configured from Advancestack Assistant (ASA)?
ASA does not allow the user to configure Spanning Tree parameters. These parameters can be changed using the switch's "console" user interface.

» Return to top

Q: What is the address table capacity of the Switch 800t?
The Switch 800t holds 10,000 MAC addresses. There is no other per port address limit

The Switch 800t uses a shared memory space for the address table for all ports. As such, there is no per port address limitation as was used in the Switch 16. The Switch 800t has a 10,000 MAC address limitation. Those 10,000 addresses can come from one port or all ports of the switch.

» Return to top

Q: What is the Broadcast Throttling feature on the Switch 800t?
Under "Automatic Broadcast Control" and "port configuration" there's a parameter called "broadcast limit" with values from 0 to 99, that "represents maximum percentage of broadcast/multicast words that can be transmitted."

» Return to top

Q: How many RJ-45 ports come with the switch?
Four UTP port (transceivers) are shipped with the 800t. The 800t can be populated with 8 utp or 8 Fiber transceivers or any combination of the two.

» Return to top

Q: What are the different types of transceivers that can be used in the switch?
The different transceivers are:

• HP J3192C Advancestack 100Base-TX UTP Transceiver
• HP J3193B Advancestack 100Base-FX Fiber-Optic Transceiver (SC)

» Return to top

Switch 2000 and Switch 800t Layer 3 Switch FAQ

Q: I am redesigning my routed network with switches. How do I configure my switches so that end nodes on different subnets can communicate with each other?
There is no switch reconfiguration required. Instead, you need to configure each end node's Default Gateway to the end node's own IP address. This configuration item may be called "default gateway," or "ROUTE ADD NET," for example. Some network stacks may require other associated changes, such as setting "SUBNET MASK" to 0.0.0.0. Please consult the software stack vendor (e.g., Microsoft, FTP Software, etc.) for details

» Return to top

Q: Should I set my client PCs' Default Gateway to a switch's IP address?
No, setting a client's default gateway to a switch's IP address does not work. Remember, the switch is not a router, it doesn't know what to do with this type of packet and will drop it.

» Return to top

Q: How do I set the Default Gateway to send my traffic through a router?
The answer to this question depends on your setup. Below are three setups. Choose the one that matches your network.

1) Same subnet. Two clients are on the same IP subnet (and the same physical segment). No Default Gateway configuration is needed. Each client ARPs for the other, then sends packets to the other client's MAC address.

2) Router (Simple case). Two clients are separated by the router. Each client has a gateway configured. That gateway address is the router's address (the port of the router closest to the client). When client1 on one subnet wants to communicate with client2 on the other subnet, client1 ARPs for the router, then sends all packets designated for client2 to the router's MAC address (and client2's IP address).

3) Router is multinetted. A router is multinetted when two clients are on the same physical segment, but their IP addresses are on different subnets or networks. The router port closest to these clients has two IP addresses; one address on client1's subnet, the other address is on client2's subnet. With a multinet setup, choose one of the three cases below.

3a) Each client has a gateway configured. That gateway address is the appropriate router's address (i.e., the router's address that is on the same subnet as the client.)

Example: Router port is 10.10.10.10 and 11.11.11.11
client1 is 10.10.10.50, default gateway 10.10.10.10
client2 is 11.11.11.50, default gateway 11.11.11.11

When client1 wants to communicate with client2, client1 ARPs for the router (10.10.10.10), then sends all packets destined for client2 to the router's MAC address (and client2's IP address). This is the same operation as in case (2), but because both clients are on the same physical segment, the packet from client1 goes to the router, then right back out the same router port to client2. Note: this causes extra traffic: two packets on the wire instead of one, compared with (3b).

3b) Client1 has its own IP address configured as default gateway and client2 still has the router's address as gateway.

Example: client1 is 10.10.10.50, default gateway 10.10.10.50
client2 is 11.11.11.50, default gateway 11.11.11.11

When client1 wants to communicate with client2, client1 would expect to send the packet to the gateway. But the gateway is itself! So client1 simply ARPs for client2 (in effect, the packet came from the gateway). Client2 receives the packet, realizes that it came from a different subnet, and sends the reply to its configured gateway (the router). The router sends that packet right back out that same router port to client1 (after changing the source and destination MAC addresses). This gives us a triangular communication path:

client1 ---> client2 ---> router ---> client1

3c) Each client has its own IP address configured as a default gateway. Now the communication from client1 to client2 is a direct path, with no router involved. This is the preferred situation when combining multiple subnets or networks onto a single non-routed segment. What if the router has Proxy ARP = YES? Client1 will receive two ARP Replies, one from the router and one from client2.

» Return to top

Q: How do the Switch 800t and Switch 2000 respond to proxy-ARP requests by using Layer 3 switching?
With ABC (Automatic Broadcast Control) on, the Switch 800t and Switch 2000 process ARP packets in firmware, while hardware typically handles packet forwarding. These switches store learned ARP information in an ARP table. This technical tip describes how the proxy-ARP requests are processed.

Let's review the steps the switches take to forward packets:

1. Switch receives a packet.
2. Switch looks up the packet's MAC source address in the address (forwarding) table.
3. If the address is not in the address table, the switch firmware creates an address table entry for it.
4. Switch looks up the packet's MAC destination address in the address table.
5. If the address is in the table, the switch hardware forwards or discards the packet depending upon the location information in the table. If the address is not in the table, the switch forwards the packet out all ports. If the address is multicast or broadcast, the switch delivers the packet to the firmware and forwards it out all ports.

On receiving an ARP request, the Switch uses the source's IP and MAC addresses to create an ARP table entry. An ARP request contains the target's IP address, but not its MAC address, so the switch cannot construct a complete ARP table entry at this time.

The switch completes the target's ARP table entry by using these methods:

1. Wait for the target to send out an ARP request (this request contains all information the ABC firmware needs to complete the ARP table entry).
2. Listen for the ARP reply from the target. Note: If the reply's destination MAC address is in the address table, the switch's hardware will forward the packet and the firmware will not see the packet. Because the firmware can't see the packet, the entry cannot be completed at this time.
3. The switch could send out its own ARP request for that target. Because the reply will be addressed to the switch itself, the firmware will read the packet.

Let's describe these methods in detail:

Method Number 1. Source1 sends an ARP request for target1 and the switch does not yet have target1 in its address table. Here's the sequence of events:

1. Switch receives (broadcast) ARP request.
2. Switch sends that packet to the firmware. If the switch doesn't have an ARP table entry for target1's IP address, the switch creates one. This entry is incomplete, because the switch doesn't know target1's MAC address.
3. Switch forwards the ARP request, unmodified (the switch's address will NOT be in the packet), out all ports except the one on which it was received.
4. Switch receives unicast ARP reply from target1. Because the switch doesn't have target1's MAC address in its address table, the switch sends a "learn interrupt" to the firmware. This gives the firmware a chance to look at the packet.
5. The firmware adds target1's MAC address and port # to the address table. Then it fills in target1's MAC address next to target1's IP address in the ARP table.

Method Number 2. Source1 sends an ARP request for target1, but the switch already has target1's address in its address table. If this occurs, the following steps are executed:

1. Switch receives broadcast ARP request.
2. Switch sends that packet to firmware, which creates a new target1 entry in the ARP table.
3. This entry is incomplete, lacking target1's MAC address.
4. Switch forwards the packet, unmodified, out all ports except the one on which it was received.
5. Switch receives unicast ARP reply from target1. But this time, since the switch already has target1's MAC address in its address table, there is no "learn interrupt". Instead, the hardware quickly finds the destination port and forwards the packet. The ABC firmware never sees the ARP reply. The ARP table is left with a partially-filled entry.

Method Number 3. is designed to handle Method Number 2 if it fails. That is, if the above method happens twice (that is, the switch hears two ARP requests for the same target, but the firmware doesn't see a reply), the switch will send ARP requests for that target out each port. These ARP requests are sent out three times, each with a different encapsulation:

1. Ethernet II
2. SNAP - with hardware type 6 (IEEE 802.3)
3. SNAP - with hardware type 1 (10 Mbps Ethernet)

Because the ARP request has the switch's MAC address, the target's reply will be a unicast reply to the switch, and the firmware will see the packet and fill in the missing MAC address in the ARP table.

Note that HP routers always ARP twice: once with Ethernet II encapsulation, and once in SNAP (IEEE 802.3) format. So even though a network analyzer will show two ARP packets for the same target, this is considered only one ARP request and will not cause the switch to initiate an ARP request as described above.

Remote end nodes (separated from the switch by a router) have ARP table entries that age after two minutes. A local end node (no router separating switch from end node) has its ARP table entry flushed after five minutes from the last time the firmware sees an IP packet from that device.

» Return to top

Q: How can I set the Default Gateway to point to the end node if it's using DHCP?
For an NT 4.0 DHCP server, set the SwitchedNetworkFlag parameter. Call Microsoft support for details.

» Return to top

Q: Does the switch reply to the end nodes' broadcast ARPs?
Only if Automatic Broadcast Control (ABC) is configured. Otherwise, the switch forwards the ARPs as normal traffic.

» Return to top

Q: How do I configure the switch to know about the different subnet addresses and their physical locations?
There is no need to configure this. The switch learns MAC addresses automatically.

» Return to top

Q: Is 10/100Mbps auto-negotiation the same as Plug-n-Play?
No. The following configuration will cause severe network problems:

The hub, switch, or router will correctly sense (not auto-negotiate) the 10Mbps or 100Mbps speed. Since the end node was configured for a specific speed and duplex state, and therefore does not negotiate, the hub, switch, or router will choose the communication mode specified by the 802.3u standard, namely half-duplex.

With one device running at half-duplex and the device on the other end of the connection at full-duplex, the connection will work reasonably well at low levels of traffic. At high levels of traffic the full-duplex device (end node, in this case) will experience an abnormally high level of CRC or alignment errors. The end users usually describe this situation as, "Performance seems to be approximately 1Mbps!". Often, end nodes will drop connections to their servers.

In this same situation, the half-duplex device will experience an abnormally high level of late collisions.

The network administrator must take care to verify the configuration of each network device during installation. Also, check the operational mode of each network device. That is, check both how you configured it and also that it comes up as you expect, for example, at 10Mbps/half-duplex.

» Return to top

Q: What is the recommended way to connect multiple VLANs between a routing switch and a layer 2 switch?
The diagram below illustrates the question.

The following HP switches provide VLANs and have a single MAC/Ethernet address (filtering) table: Switch 800t, 2000, 1600m, 2400m, 2424m, 4000m, 8000m. In the diagram above we show a Switch 8000m, but the following discussion applies to all of the switches listed in the previous sentence. The HP ProCurve Switch Routing Switch 9304m, 9308m, or 6308m-SX, as a default gateway, has a single MAC address (for all of its VLANs) if using virtual Ethernet interfaces. In the diagram above we show a 9304m, but this could be a 9308m or 6308m-SX as well.

Let's consider PC "A" attempting to send an IP packet to PC "B". PC "A" will send the 8000m a packet with the 9304m's MAC address in the destination field. If the 8000m has not yet learned this MAC address, the 8000m will flood the packet out all of its VLAN1 ports, including the VLAN1 link to the 9304m. The 9304m will then route the packet toward PC "B" via its link with the 8000m's VLAN2 connection. The 8000m will enter the 9304m's MAC address into its MAC address table as located in VLAN2. The 8000m will also forward the packet to PC "B".

Let's consider a second packet that PC "A" sends to PC "B". PC "A" sends the packet, again addressed to the 9304m's MAC address, to the 8000m. The 8000m will check its address table and find that the 9304m appears to be located on VLAN2. Since the 8000m believes that this MAC address is not located on VLAN1, the switch will discard the packet.

Later, when the 9304m transmits a packet to the 8000m via the VLAN1 link, the 8000m will update its address table to indicate that the 9304m's MAC address is located in VLAN1 instead of VLAN2. As you can see, the 8000m's location information for the 9304m's MAC address will vary over time between VLAN1 and VLAN2. For this reason, some packets directed through the 8000m for the 9304m's MAC address will be discarded. Performance may appear to be poor or connectivity may appear to be broken.

To avoid this issue, simply use one cable between the 8000m and the 9304m instead of two, making sure that the two VLANs use tags on that link, as shown below.

» Return to top

Q: Why is my Macintosh system unable to use AppleTalk services?
Possible symptoms include: no AppleTalk services, only local network AppleTalk services, performance problems, and intermittent network services. If you remove the Macintosh from its dedicated switch (or routing switch or router) port and connect it to a hub, the problem goes away.

If the switch (or routing switch or router) has Spanning Tree Protocol enabled, see Apple Computer, Inc's Tech Info Library entry "Spanning Tree Protocol: AppleTalk Issues".

» Return to top

Q: Why does my RS-232 console cable not work with certain HP Advancestack and switches?
There are many possible sources of console connection problems, including terminal emulator defects and incorrect configurations. Your HP Advancestack or HP ProCurve Switch installation documentation discusses the correct configuration of the console port.

There is a known problem with a particular RS-232 cable that shipped with some of the following products:

• J3178A HP Advancestack Switch 208/224 Management Module
• J3245A HP Advancestack Switch 800t
• J3100A HP Advancestack Switch 2000 (version A)
• J3100B HP Advancestack Switch 2000 (version B)

If the RS-232 cable with HP part number 5183-7247 is attached to the console port of certain HP Advancestack switch models before the switch is reset or powered up, the console will display:

Decompressing...done

then give no more output. You will not be able to access the switch's user interface through the console.

If the 5183-7247 RS-232 cable is attached to the switch's console port after the switch is already operational, you will not be able to see or access the switch's user interface.

The 5183-7247 cable cannot be made to work properly with the switches listed above at any baud rate or with any terminal emulator. The reason is that this cable does not transmit certain RS-232 signals required by those switches.

There is no problem with the RS-232 cable part number 5182-4794 that shipped earlier with the four Advancestack devices listed above.

If you have the cable 5183-7247 that you are trying to use with one of the HP Advancestack devices listed above, please contact your local HP Customer Care Center, which can arrange for a replacement cable to be sent to you at no charge as part of your product's warranty.

» Return to top

Q: Why won't my Xircom Realport-REM56G-100BTX Notebook NIC operate properly with my HP ProCurve Switch switch in full duplex mode at 100Mbps?
HP knows that this Xircom NIC, at full duplex over 100Mbps, does not operate with our HP ProCurve Switch switches. HP has no reports of any other vendors' NIC having a similar issue with our ProCurve switches. While this NIC's data sheet states support of full duplex, HP has not been able to get this NIC to operate at full duplex over a 100Mbps connection, regardless whether it is configured to Auto-negotiate or full duplex, fixed configuration.

We are not aware of any fixes for this issue.

Failure symptoms include poor performance and lost connections. See the discussion of "Q: Is 10/100Mbps auto-negotiation the same as Plug-n-Play?" on this same page for more details on the problem symptoms.

The workaround is to operate the NIC in half-duplex mode or to use a different NIC.

» Return to top