GIGAswitch/IP Version 2.0.1-D Release Notes
Open Problems

We here at Digital Equipment Corporation believe very strongly in providing you with any information that may be useful in the maintenance and support of your DIGITAL equipment. To this end, we have put into motion a series of initiatives aimed at leveraging the collective experiences of other users and internal personnel in the resolution of networking problems. As one example and a demonstration of this commitment, we have provided in this chapter problem summaries taken directly from our problem database in their "raw" form.

Please note that these problems have not yet been acknowledged or verified as defects in DIGITAL equipment in any manner. They represent work in progress only. Some of these problems may be erroneous or eventually be isolated to networking problems completely unrelated to DIGITAL equipment.

They are being presented here in a spirit of cooperation and openness with the belief that arming you with some information will be more useful than providing you with no insight at all into potential issues. Also, we would be very interested to hear about information you may have regarding these open reports. If you have any information that will be useful in isolating these issues or have any questions about a specific problem please do not hesitate to contact DIGITAL Multivendor Customer Service or your DIGITAL authorized value-added reseller (VAR) or distributor.

We will continue to work diligently at improving the quality of this information and are committed to providing you with high quality, reliable networking products.

This section describes known problems and workarounds.

	IGRP is missing from the Web configuration page.
Using the following patch, you can see IGRP in the active file. To permanently (and forcibly) enable IGRP on a GIGAswitch/IP Version 2.0.1-D product, complete the following steps: Log in as admin. Enter the following commands: # cd /usr/local/config/bin # ./dbset "protocol:igrp" t # ./dbset save save
	The IPX Protocol is not supported in Version 2.0.1-D of the GIGAswitch/IP Software as advertised. The IPX Protocol will be supported in Version 2.1-D.
	The vi editor exits with an error and freezes the terminal session.
When using the vi editor on a serial console, the program sometimes exits with the following message: endwin inappropriate ioctl for device tty freezup This causes the tty to freeze up. The only way to restart it is to Telnet in and kill the immediate child of getty, thus initiating a new login.
	RFC 1812 violation--broadcast is forwarded.
Using ping to a host not on the local subnet using the link layer broadcast address should not work.
	netstat generates a spurious error message: netstat: kvm_read: kvm_read: Bad address.
When the utility is run during a route flap, it cannot cope with information changing while it is running. Simply re-run the program and it should function normally.
	Under Mosaic 2.7b4, Monitor Tool Switch View links do not work.
	SNMP returns ifType as fibreChannel(56) and ipSwitch(78) when it should return ifType as Ethernet(6) and SONET(39).
This may be fixed in future releases.
	It is not possible to create a QoS rule using the Lynx interface.
When Lynx is used to create a new rule, even if inbound and outbound options are checked, an error is returned stating that the rule is not specified.
	Field error check messages don't tell you which parameter is incorrect.
On screens where you may enter multiple entries on the same commit, the error checking message does not tell you which field is actually invalid.
	SNMP agents on corporate network are showing invalid routes.
All of the corporate network routers are showing invalid routes in MIB-II. These routes do not show up in netstat.
	Panic in driver_control_input on controller.
	Traversal of clearVISN IP Switch Manager fails due to recursive URL.
When you turn on help from the rate shaping stats page and then select the events button, continuing to select the events button generates a new URL each time. The traversal never ends.
	Possible multicast tunnel corruption when packet size is greater than the MTU.
Problem trying to use multicast tunnels with precept software. In native multicast mode, the application works fine. However, if the multicast UDP packets are tunneled, the application does not work. The originating application is sending UDP packets of size 1452. The packets coming out of the tunnel appear to have some strange fragment values.
	Number of IFMP clients not displayed when configured during full install.
If you do a full installation on a unit and specify a number of IFMP clients, this will not be displayed in clearVISN IP Switch Manager when you view the license information
	Panic in m_copym running multicast.
Panic appears when sending a multicast packet with size=200 using Ipsilon multicast ping and when running OSPF with anv1 simulating 100 neighbors. ping -s 2000 -j 224.1.2.3
	The current status of Ethernet is not checked when sending routing updates.
If the port cable is disconnected, the system still thinks the Ethernet port is up.
	Simple authentication for VRRP produces parse error.
	Entering an illegal cluster value of 257 for VRRP still shows the interface page with that value on the GUI.
Entering an illegal value in clearVISN IP Switch Manager does not enter it in the system database, but it still shows the interface page to be configured with that illegal cluster value in clearVISN IP Switch Manager.
	Multicast forwarding may not work in certain circumstances.
	Rate shape rule corruption in clearVISN IP Switch Manager (Name field).
This was not reproducible.
	An EAR interface turned off administratively still allows multicast traffic to pass through.
	Upgrade from 1.3 to 2.0 does not convert gateway/controller in license page.
	Handling of deletes causes switch to go into a delete responding loop session and never recover.
	IP identifier byte swapped in fragments.
The ip_id byte swapped on the first fragment.
	IP fragments reordered during forwarding.
In sending some IP packets that are sent as two fragments, on the incoming interface, the packets consistently show up with the first fragment first, and the second fragment second. On the outgoing interface, the fragments are re-ordered.
	Certain characters have problems with rate shape rules (# and &).
The # character will cause all characters to the right to not be displayed. The & character will cause the entire name to not be recognized.
	Unable to do full installation on half-duplex, 100 Mb/s Ethernet link.
Attempted to do a full installation on a 100 Mb/s Ethernet card with half-duplex setting on specific NICs. Fails while trying to do the initial FTP session.
	Initial attempt of a full installation using some types of Ethernet NICs fail for full-duplex, 100 Mb/s link.
Attempted to do a full installation using the 100 Mb/s option and the full-duplex mode. Initially it failed, but the next attempt succeeded.
	ifaddr added to system list uninitialized at low spl.
Difficult to reproduce. This showed up when running VRRP (i.e., every interface has 2 IP addresses) on a fully populated Ipsilon FAS1200.
	Controller goes into a constant reboot cycle.
Switch controller is in a constant reboot cycle and never comes out of it. A power cycle seems to help. This has happened once in the past.
	Single flow on one EAR port doesn't get switched.
A 50,000 packets per second (128 byte) data stream was set up on the second port of a FAS card, destined for a host reached through another blade. This flow was never switched; the incoming and outgoing states of the flow remained at ifmpc. This caused the controller to remain completely at interrupt level. When setting up another 50,000 packets per second (128 byte) data stream from a different source, to the same destination, through the first port on the same FAS card, that stream was switched fine. Furthermore, once that stream was running, starting the original data stream on the first port caused it to be switched as well. The problem showed up only when the second port was used alone.
	IPX shows ON for all interfaces when clearVISN IP Switch Manager has it displayed off.
	xpand is leaking memory.
	IPX SAP GNS cannot be disabled through clearVISN IP Switch Manager or Lynx.
	"Manual remote address" when disabled (from an enabled state) still retains all info.
	IP Switch software Gateway to 3Com Switched Media hub (Linkbuilder 1000) performance is slow.
	When setting RIP v1 broadcast a syntax error is produced.
	No VRRP information displays on the Configuration Summary page.
	Fragmented packets cause the crc_error, sml_fbq_empty, and max_len_err to go up constantly.
Fragmented packets causes the following counters to go up on a controller: rx_crc_errors rcv_max_len_err sml_fbq_empty Traffic is as follows: host ->ear->controller->gateway->host VRRP is configured on the EAR interface. The ATM NIC on the controller is an IDT NIC (Rev E).
	vmstat -m doesn't work.
	In certain rare instances, rate shaping rules don't take effect, even with the apply button.
Customers will rarely, if ever, want to disable switching, so this probably has low impact.
	DIGITAL GIGAswitch/IP software forwards broadcast packets as multicast on the same interface.
	Panic in ifmpc_add_shaped_defaults.
This was caused by a race condition where IFMP-C sync was lost while a pending GSMP message for the shaped defaults was being added.
	IPSRD is dumping core during generation of /etc/ipsrd.conf.
Not reproducible.
	Turning Bootp relay enable *off* does not stop the bootpgw process.
	IPSRD core dumps:task_block_sbrk: sbrk(4096): Cannot allocate memory.
	At times under flows in Monitor tool inaccurate numbers (much too high) are displayed.
	Upgrade from 1.3 to 2.0 causes problems with RIP to OSPF configuration.
When a system is upgraded from 1.3 to 2.0 and there is RIP to OSPF exports configured the active file will become corrupted and prevent the user from making any changes at all to the RIP portion of the unit. The clearVISN IP Switch Manager displays the following message: When exporting routes into rip, a metric value must be set.
	No entry for VRRP in /etc/ipsrd.conf even though it is in the active db file.
Configured VRRP using the new syntax after removing the old VRRP lines in the db file. There were no syntax errors but the IPSRD.conf doesn't contain any VRRP entries even after a commit.
	VRRP committer doesn't complain if cluster is not specified.

---

---