From Ronald.vanderPol@rvdp.org Wed Jun 4 14:27:14 2003 From: Ronald.vanderPol@rvdp.org (Ronald van der Pol) Date: Wed, 4 Jun 2003 15:27:14 +0200 Subject: [Xorp-users] list archives not available? Message-ID: <20030604132714.GA3606@rvdp.org> The archive on http://mailman.icsi.berkeley.edu/pipermail/xorp-users/ does not seem to work. Actually, some other archives have the same problem. It seems that the latest message I received on xorp-users was on on March 27. It that correct? I would like to keep in touch with the project. On http://www.xorp.org/news.html I see two news items in May. Were these sent to one of the mailing lists too? If so, which one? Currently, I am only subscribed to xorp-users (althought I am not using xorp yet :-) rvdp From adam@hiddennet.net Wed Jun 4 16:20:44 2003 From: adam@hiddennet.net (Adam Greenhalgh) Date: 04 Jun 2003 16:20:44 +0100 Subject: [Xorp-users] list archives not available? In-Reply-To: <20030604132714.GA3606@rvdp.org> References: <20030604132714.GA3606@rvdp.org> Message-ID: <1054740044.12134.93.camel@localhost> Ronald, Xorp users is fairly inactive, and that was the last email on xorp users I receieved. This link works http://mailman.icsi.berkeley.edu/mailman/listinfo/xorp-users Neither of the items on the new page were sent to the mailing lists, internal or external. Xorp is gearing up for a 0.3 release in the near future so keep an eye on xorp-users and on the web page. Hope this helps. Adam On Wed, 2003-06-04 at 14:27, Ronald van der Pol wrote: > The archive on http://mailman.icsi.berkeley.edu/pipermail/xorp-users/ > does not seem to work. Actually, some other archives have the same > problem. > > It seems that the latest message I received on xorp-users was on on > March 27. It that correct? > > I would like to keep in touch with the project. On > http://www.xorp.org/news.html I see two news items in May. Were > these sent to one of the mailing lists too? If so, which one? > Currently, I am only subscribed to xorp-users (althought I am not > using xorp yet :-) > > rvdp > _______________________________________________ > Xorp-users mailing list > Xorp-users@xorp.org > http://mailman.ICSI.Berkeley.EDU/mailman/listinfo/xorp-users From mjh@icir.org Wed Jun 4 18:42:49 2003 From: mjh@icir.org (Mark Handley) Date: Wed, 04 Jun 2003 10:42:49 -0700 Subject: [Xorp-users] list archives not available? In-Reply-To: Your message of "04 Jun 2003 16:20:44 BST." <1054740044.12134.93.camel@localhost> Message-ID: <23540.1054748569@vulture.icir.org> >Xorp users is fairly inactive, and that was the last email on xorp users >I receieved. This link works >http://mailman.icsi.berkeley.edu/mailman/listinfo/xorp-users > >Neither of the items on the new page were sent to the mailing lists, >internal or external. Xorp is gearing up for a 0.3 release in the near >future so keep an eye on xorp-users and on the web page. As the code gets more stable and functional and as more people get interested, we're going to have to do a better job of keeping people informed. Perhaps we should send out monthly progress reports? Let us know what you want. Cheers, Mark From Ronald.vanderPol@rvdp.org Wed Jun 4 21:50:31 2003 From: Ronald.vanderPol@rvdp.org (Ronald van der Pol) Date: Wed, 4 Jun 2003 22:50:31 +0200 Subject: [Xorp-users] list archives not available? In-Reply-To: <1054740044.12134.93.camel@localhost> References: <20030604132714.GA3606@rvdp.org> <1054740044.12134.93.camel@localhost> Message-ID: <20030604205031.GB11073@rvdp.org> On Wed, Jun 04, 2003 at 16:20:44 +0100, Adam Greenhalgh wrote: > Ronald, > > Xorp users is fairly inactive, and that was the last email on xorp users > I receieved. This link works > http://mailman.icsi.berkeley.edu/mailman/listinfo/xorp-users Are you sure? If I click on the "Xorp-users Archives" link I get a "The requested URL /pipermail/xorp-users/ was not found on server mailman.icsi.berkeley.edu" error. rvdp From Ronald.vanderPol@rvdp.org Wed Jun 4 21:56:43 2003 From: Ronald.vanderPol@rvdp.org (Ronald van der Pol) Date: Wed, 4 Jun 2003 22:56:43 +0200 Subject: [Xorp-users] list archives not available? In-Reply-To: <23540.1054748569@vulture.icir.org> References: <1054740044.12134.93.camel@localhost> <23540.1054748569@vulture.icir.org> Message-ID: <20030604205643.GC11073@rvdp.org> On Wed, Jun 04, 2003 at 10:42:49 -0700, Mark Handley wrote: > As the code gets more stable and functional and as more people get > interested, we're going to have to do a better job of keeping people > informed. Perhaps we should send out monthly progress reports? Let > us know what you want. Monthly reports sounds like a good idea to me. BTW, I just read on your web site that the xorp-users archive is inaccessible at the moment. Sorry for my previous emails :-) rvdp From jef@linuxbe.org Thu Jun 5 09:52:27 2003 From: jef@linuxbe.org (Jean-Francois Dive) Date: Thu, 5 Jun 2003 10:52:27 +0200 Subject: [Xorp-users] list archives not available? In-Reply-To: <23540.1054748569@vulture.icir.org> References: <1054740044.12134.93.camel@localhost> <23540.1054748569@vulture.icir.org> Message-ID: <20030605085227.GB1596@gardafou.assamite.eu.org> Hi all, hey some activity on this ML;) Well as far as i can see, this project seems very interesting but there is no much visibility from the outside world. The framework seems very good to me, i haven't dig the code much, but i would potentially interested to see this further. I have some general questions on the way you the project going: - how well is the C++ code you write fit the embedded world (as i said i haven't look at the code, but i would not like to see too much of run-time operations (dyncasts for expl), as well as exceptions as well as too much encapsulation which lead to too much dead code (expl: only use pointers and no references to avoid to have too much copy constr. etc..) ?? - I think the project could fill a gap in what's on the open source market now: a real router environement that could be the core of a router distro (based on linux or BSD or whatever), as you state that the project's main interest is to have some open and deployed infrastructure for network proto testing and improvement. Where do you see the boundaries of the project, or what's the main type of router you want to see developped ? (i just say that because routing protocols seems to be the central point around the developement which is for exemple useless in the most classical open source router aka gateways..). Cheersm JeF On Wed, Jun 04, 2003 at 10:42:49AM -0700, Mark Handley wrote: > > >Xorp users is fairly inactive, and that was the last email on xorp users > >I receieved. This link works > >http://mailman.icsi.berkeley.edu/mailman/listinfo/xorp-users > > > >Neither of the items on the new page were sent to the mailing lists, > >internal or external. Xorp is gearing up for a 0.3 release in the near > >future so keep an eye on xorp-users and on the web page. > > As the code gets more stable and functional and as more people get > interested, we're going to have to do a better job of keeping people > informed. Perhaps we should send out monthly progress reports? Let > us know what you want. > > Cheers, > Mark > _______________________________________________ > Xorp-users mailing list > Xorp-users@xorp.org > http://mailman.ICSI.Berkeley.EDU/mailman/listinfo/xorp-users -- -> Jean-Francois Dive --> jef@linuxbe.org There is no such thing as randomness. Only order of infinite complexity. - Marquis de LaPlace - deterministic Principles - From mjh@icir.org Mon Jun 9 23:31:18 2003 From: mjh@icir.org (Mark Handley) Date: Mon, 09 Jun 2003 15:31:18 -0700 Subject: [Xorp-users] Announcing XORP Release 0.3 Message-ID: <92282.1055197878@aardvark.icir.org> On behalf of the entire XORP team, I'm delighted to be able to announce XORP release 0.3, now available from www.xorp.org. Highlights since our last release in March include: - Significantly improved inter-process integration and error handling. - Many improvements to the XRL IPC mechanism, especially related to monitoring process start and termination. - Started work on an SNMP interface to XORP, with partial support for the BGP4 MIB now implemented and working. - The multicast FEA (MFEA) has been merged with the FEA, providing a more consistent and integrated interface with the forwarding engine. - RIB support for providing the full routing table to other processes such as PIM-SM. - PIM Sparse-Mode and MLD/IGMP are now reasonably complete and fairly well tested. - Work has started and on our RIP implementation. Full release notes are included below. In general, work is progressing very well, although we would not yet recommend running XORP in production routing environments. However, as the architecture is basically complete though, this release of XORP might be useful as the basis for research. If you're interested in the cutting edge, the current snapshot is always available by anonymous CVS. This release indicates a stable point where the documentation matches the code and all the isolation tests are known to succeed on our supported platforms. As always we'd welcome your comments - xorp-users@xorp.org is the right place for general discussion, and private feedback to the XORP core team can be sent to feedback@xorp.org. - The XORP Team ------------------------------------------------------------------ XORP RELEASE NOTES Release 0.3 (2003/06/09) ======================== ALL: - Compilation fixes for GCC3.3 on FreeBSD 4.8R. - Cross directory build and gmake check support. - Purge of abort() calls. Replacement by XLOG_UNFINISHED, XLOG_UNREACHABLE and XLOG_ASSERT as appropriate, or in some cases better error handling code. - Change to autoconf-2.53 and automake-1.5 as standard autotool versions. The older autoconf-2.13 and automake-1.4 should still work. - Remove the multicast-specific timer implementation, and use instead the default Xorp timers (XorpTimer). LIBXORP: - Allocation without construction bug fixed in SelectorList code. Fixes potential for junk values and sanity checking when the number of active file descriptors is large. - Addition of status codes for XORP processes to report operational state (to the rtrmgr). - Addition of XLOG_UNFINISHED and XLOG_UNREACHABLE to xlog.h. - Explicitly disable copy constructor and assignment operator for EventLoop. - Replacement of traditional struct timeval with TimeVal class. TimeVal comes with constructors and numerous helper methods. - IPvXNet methods renaming: get_ipv4Net() -> get_ipv4net() get_ipv6Net() -> get_ipv6net() LIBXIPC: - Removal of 0.1 release Finder code and compatability shims that allowed newer and older code to co-exist. Renaming of "FinderNG" classes and files to "Finder". - Use of class and instance names and Finder policy for having a primary instance in a class (first registered is primary, then second in case first exits/fails, and so forth). - Support for tunneling XRLs and XRL responses to and from the Finder. - Implementation event notification through the Finder using tunneled XRLs. XRL Targets can now request to be notified of the birth and death of other XRL Targets by the Finder. - Reduced number of XrlErrors to hopefully simplify handling errors. - Addition of new enumated type XrlErrorCode and change XrlError::error_code method to return this type. Allows compiler to detect unhandled error types in switch statements (if default case is not used). - Finder clients no longer reconnect to the Finder should they lose their connection per ${XORP}/docs/design_arch/error_handling.tex. - Increased Finder timeout for call_xrl. - Fix memory corruption following Finder transport error events. - Convert all code to use FinderServer class when they need to instantiate a Finder. - More consistent handling of defaults on send and receive sides of Finder. - Fix xrl_parser_input.cc handling of absolute file names. - Assorted minor clean-ups. XRL: - Addition of get_status and shutdown methods to the common interface. RTRMGR: - Workaround for yacc incompatibilities on newer Redhat releases. - Implement a new TaskManager, which handles actually doing the work during a commit, starting processes and validating that they start, dependency interlocking between processes, and shutting down processes that are no longer needed. - rtrmgr will now shutdown processes after they are not needed. - rtrmgr can now use the common XRL call get_status to monitor process state, and the common XRL shutdown to gracefully shutdown a process. Added new modinfo subcommands to specify when to use XRLs for these tasks. - Use system MD5 for authentication. - Better config file string parser. - Add tinderbox tests to cover some of the basic functionality. - Better error handling (more compliant with XORP Error Handling document). XORPSH: - Ctrl-C now terminates the current command, Ctrl-D exits xorpsh. FEA: - MFEA code merged with the unicast FEA, though logically the MFEA is still a separate entity. - Refactoring of FEA internals regarding reading/writing/observing interfaces-related information and forwarding-table related information. - Support for few more OS-specific methods for reading interface-related information or forwarding table related information. - Addition of IPv6 support (where missing) for reading/writing information regarding network interfaces or the unicast forwarding table. - Addition of a command-line option to ``fea'' and ``fea_dummy'' to specify the finder hostname and port number. - On startup and shutdown remove from the kernel all forwarding entries added by the fea. - Addition of XRLs to obtain information about all network interfaces (i.e., not only those that are configured by the rtrmgr). Should be used only for debugging purpose (though temporary they are used by the MFEA as well). - ``fea_rtsock'' deprecated (replaced with the generic ``fea''). RIB: - Export a RIB table to any target that registers interest. - Replace fea_fti/0.1 XRL interface with fti/0.2. The new interface introduces metric, admin_distance and protocol_origin when RIB information is propagated to the RIB clients. - Refactoring of the Vif manager implementation. - Support for returning the process status through thet get_status XRL and shutting down using the shutdown XRL. BGP: - More robust to XRL communication failures. In particular support for retransmission. - Replace calls to abort() with more self-explanatory self-destruction. - Terminate interface removed and replaced with common shutdown XRL interface. - Use system md5 header files. - Peer list sorted by remote IP for snmp. - Uninitialised variable bug fix in NextHopRibRequest. - Initial support for returning the process status via get_status XRL. MFEA: - MFEA code merged with the unicast FEA, though logically the MFEA is still a separate entity. As part of the merging process, the MFEA now obtains the information about network interfaces and the unicast forwarding table from the FEA instead of the the kernel. - Addition of a check for Router Alert IP option. - Enable receiving of hop-by-hop options. - Replace all MLD6_XXX with MLD_XXX (following KAME's recommendation). - Compilation fixes for NetBSD-current (and probably NetBSD-1.6.1) (courtesy Hitoshi Asaeda ). - Initial support for returning the process status via get_status XRL. - Addition of a command-line option to ``test_mfea'' to specify the finder hostname and port number. - Misc. bug fixes, internal code refactoring, and cleanup. MLD/IGMP: - Protocol implementation testing completed. - Replace all MLD6_XXX with MLD_XXX (following KAME's recommendation). - Unify some of the MLD and IGMP protocol-specific code. - Support for configurable setting of the protocol version per vif. - Addition of a command-line option to ``test_mld6igmp'' to specify the finder hostname and port number. - Misc. bug fixes. PIM-SM: - Configuration-related fixes. - Rename RP-related configuration commands and XRLs: OLD: add_config_rp{4,6}, delete_config_rp{4,6}, config_rp_done NEW: add_config_static_rp{4,6}, delete_config_static_rp{4,6}, config_static_rp_done - Addition of XRL interface to receive MRIB-related updates from the RIB; the interface is not used yet. - Initial support for returning the process status via get_status XRL. - Addition of a mechanism that should hold-off PIM-SM status to PROC_NOT_READY until it has received all necessary confirmation from both MFEA and MLD/IGMP. - Addition of PIM configuration option and XRL interface to configure the SPT-switch bandwidth threshold. Currently, the SPT-switch is enabled only for the corner cases: switch on the first packet, or never switch. - Addition of a command-line option to ``test_pim'' to specify the finder hostname and port number. CLI: - Critical bug fix (could be triggered if a program like ${XORP}/pim/test_pim is run in background, and is accessed by the built-in CLI). - Addition of Ctrl-C handling: o If Ctrl-C is hit before , then cancel the current line and present a new prompt. o If Ctrl-C is hit while waiting for command completion, then cancel the wait and present a new prompt. - Removed the 5 seconds timer that would timeout the waiting for command completion (the user can use Ctrl-C to cancel the wait). - Enable receiving Ctrl-C if the input is stdio (e.g., in case of xorpsh). SNMP: - Integrated with Net-SNMP agent. - Implemented bgpVersion and bgpPeerTable from BGP4-MIB (RFC 1657). RIP: - Implementation started. From ypol@intracom.gr Tue Jun 17 12:37:22 2003 From: ypol@intracom.gr (Polychronis Ypodimatopoulos) Date: Tue, 17 Jun 2003 14:37:22 +0300 Subject: [Xorp-users] newbie in XORP Message-ID: <3EEEFD72.4040000@intracom.gr> Hello XORP installed ok, but when I try to run xorpsh it gives me this: step3 (fea) step3: execute TaskXrlItem::execute xrl fea/ifmgr/0.1/commit_transaction?tid:u32=$(interfaces.TID) +++++++[ 2003/06/17 14:02:08 ERROR fea:19666 FEA +382 ifconfig_set_ioctl.cc push_config ] On rl0: O/S does not recognise interface [ 2003/06/17 14:02:08 WARNING fea XrlFeaTarget ] Handling method for ifmgr/0.1/commit_transaction failed: XrlCmdError 102 Command failed On rl0: O/S does not recognise interface TaskXrlItem::execute_done xrl_done (fea) TaskManager::task_done task failed ############################################################## ## commit_pass2_done ## commit failed: 102 Command failed On rl0: O/S does not recognise interface MasterConfigTree::config_done: fail: 102 Command failed On rl0: O/S does not recognise interface [ 2003/06/17 14:02:08 ERROR rtrmgr:19665 RTRMGR +127 master_conf_tree.cc config_done ] Startup failed (102 Command failed On rl0: O/S does not recognise interface) Startup failed (102 Command failed On rl0: O/S does not recognise interface) Obviously I am doing something wrong but, what? My actual goal is to test MLD on Linux. Are there any shortcuts to that? Thank you very much. Polychronis Ypodimatopoulos INTRACOM S.A. From greenhal@icir.org Tue Jun 17 12:48:43 2003 From: greenhal@icir.org (Adam Greenhalgh) Date: 17 Jun 2003 12:48:43 +0100 Subject: [Xorp-users] newbie in XORP In-Reply-To: <3EEEFD72.4040000@intracom.gr> References: <3EEEFD72.4040000@intracom.gr> Message-ID: <1055850523.24834.34.camel@localhost> Hi Part of your problem, I think, is that you haven't modified rtrmgr/config.boot to reflect the linux interface names, rl0 is a freebsd network interface name, whereas under linux it is likely to be eth0. Sorry what is MLD ? Adam On Tue, 2003-06-17 at 12:37, Polychronis Ypodimatopoulos wrote: > Hello > > XORP installed ok, but when I try to run xorpsh it gives me this: > > step3 (fea) > step3: execute > TaskXrlItem::execute > xrl fea/ifmgr/0.1/commit_transaction?tid:u32=$(interfaces.TID) > +++++++[ 2003/06/17 14:02:08 ERROR fea:19666 FEA +382 > ifconfig_set_ioctl.cc push_config ] On rl0: O/S does not recognise interface > [ 2003/06/17 14:02:08 WARNING fea XrlFeaTarget ] Handling method for > ifmgr/0.1/commit_transaction failed: XrlCmdError 102 Command failed On > rl0: O/S does not recognise interface > TaskXrlItem::execute_done > xrl_done (fea) > TaskManager::task_done > task failed > ############################################################## > ## commit_pass2_done > ## commit failed: 102 Command failed On rl0: O/S does not recognise > interface > MasterConfigTree::config_done: fail: 102 Command failed On rl0: O/S does > not recognise interface > [ 2003/06/17 14:02:08 ERROR rtrmgr:19665 RTRMGR +127 > master_conf_tree.cc config_done ] Startup failed (102 Command failed On > rl0: O/S does not recognise interface) > Startup failed (102 Command failed On rl0: O/S does not recognise interface) > > Obviously I am doing something wrong but, what? > My actual goal is to test MLD on Linux. Are there any shortcuts to that? > Thank you very much. > > Polychronis Ypodimatopoulos > INTRACOM S.A. > > _______________________________________________ > Xorp-users mailing list > Xorp-users@xorp.org > http://mailman.ICSI.Berkeley.EDU/mailman/listinfo/xorp-users From ypol@intracom.gr Tue Jun 17 13:09:58 2003 From: ypol@intracom.gr (Polychronis Ypodimatopoulos) Date: Tue, 17 Jun 2003 15:09:58 +0300 Subject: [Xorp-users] newbie in XORP In-Reply-To: <1055850523.24834.34.camel@localhost> References: <3EEEFD72.4040000@intracom.gr> <1055850523.24834.34.camel@localhost> Message-ID: <3EEF0516.1060801@intracom.gr> Correct! Multi-thanks for that.
Still, I think that there should be some more descriptive manual on
how to use rtrmgr and its relation to xorpsh.
For example when I run rtrmgr a "fea" process is created.
Do I get to bring fea down through rtrmgr, or do I need to that only
through xorpsh?

When I try to run xorpsh, it gives me this:
[ 2003/06/17 15:06:14  ERROR xorpsh:20587 XifFinder +376 finder_client.cc reg_callback ] Failed to register client named xorpsh-20587-pcypol.intranet.gr-944079c1656e32585f7f67ad996e9016@127.0.0.1 of class xorpsh-20587-pcypol.intranet.gr: "211 Reply timed out"
Failed to register with router manager process
200 Finder not present/ready

So how do I shut it down?

MLD is the Multicast Listener Discovery protocol for multicast routing.
It is supposed not to have been officialy tested yet,
but I desperately need to try it on Linux.

Thank you all for your help.

P. Ypodimatopoulos

Adam Greenhalgh wrote:
Hi

Part of your problem, I think, is that you haven't modified
rtrmgr/config.boot to reflect the linux interface names, rl0 is a
freebsd network interface name, whereas under linux it is likely to be
eth0. 

Sorry what is MLD ?

Adam
 
On Tue, 2003-06-17 at 12:37, Polychronis Ypodimatopoulos wrote:
  
Hello

XORP installed ok, but when I try to run xorpsh it gives me this:

step3 (fea)
step3: execute
TaskXrlItem::execute
   xrl fea/ifmgr/0.1/commit_transaction?tid:u32=$(interfaces.TID)
+++++++[ 2003/06/17 14:02:08  ERROR fea:19666 FEA +382 
ifconfig_set_ioctl.cc push_config ] On rl0: O/S does not recognise interface
[ 2003/06/17 14:02:08 WARNING fea XrlFeaTarget ] Handling method for 
ifmgr/0.1/commit_transaction failed: XrlCmdError 102 Command failed On 
rl0: O/S does not recognise interface
TaskXrlItem::execute_done
xrl_done (fea)
TaskManager::task_done
task failed
##############################################################
## commit_pass2_done
## commit failed: 102 Command failed On rl0: O/S does not recognise 
interface
MasterConfigTree::config_done: fail: 102 Command failed On rl0: O/S does 
not recognise interface
[ 2003/06/17 14:02:08  ERROR rtrmgr:19665 RTRMGR +127 
master_conf_tree.cc config_done ] Startup failed (102 Command failed On 
rl0: O/S does not recognise interface)
Startup failed (102 Command failed On rl0: O/S does not recognise interface)

Obviously I am doing something wrong but, what?
My actual goal is to test MLD on Linux. Are there any shortcuts to that?
Thank you very much.

Polychronis Ypodimatopoulos
INTRACOM S.A.

_______________________________________________
Xorp-users mailing list
Xorp-users@xorp.org
http://mailman.ICSI.Berkeley.EDU/mailman/listinfo/xorp-users

From mjh@icir.org Tue Jun 17 17:59:28 2003 From: mjh@icir.org (Mark Handley) Date: Tue, 17 Jun 2003 09:59:28 -0700 Subject: [Xorp-users] newbie in XORP In-Reply-To: Your message of "Tue, 17 Jun 2003 15:09:58 +0300." <3EEF0516.1060801@intracom.gr> Message-ID: <92295.1055869168@vulture.icir.org> > >Correct! Multi-thanks for that.
>Still, I think that there should be some more descriptive manual on
>how to use rtrmgr and its relation to xorpsh.
>For example when I run rtrmgr a "fea" process is created.
Rtrmgr will start the processes needed to implement the configuration you specify in config.boot. So, if you specified any interface configuration, an fea will be run. Or if you specified anything that has a dependency on fea, then the fea will be run. It will be shut down when you cleanly kill the rtrmgr, or when you edit the running config using xorpsh so that there is no config that requires the fea to be running. But pretty much everything in xorp has an indirect dependency on the fea. >When I try to run xorpsh, it gives me this:
>[ 2003/06/17 15:06:14  ERROR xorpsh:20587 XifFinder +376 >finder_client.cc reg_callback ] Failed to register client named >xorpsh-20587-pcypol.intranet.gr >-944079c1656e32585f7f67ad996e9016@127.0.0.1 >of class xorpsh-20587-pcypol.intranet.gr: "211 Reply timed out"
>Failed to register with router manager process
>200 Finder not present/ready
Looks like the rtrmgr is no longer running, but has left some running processes behind. It shouldn't do this, but it still happens under some circumstances. Use kill -9 to kill the fea process. BTW, there is currently no multicast support at all via the rtrmgr. See etc/templates to see which processes currently have rtrmgr templates. This is on the list for the next release. Cheers, Mark From pavlin@icir.org Wed Jun 18 01:15:18 2003 From: pavlin@icir.org (Pavlin Radoslavov) Date: Tue, 17 Jun 2003 17:15:18 -0700 Subject: [Xorp-users] newbie in XORP In-Reply-To: Message from Polychronis Ypodimatopoulos of "Tue, 17 Jun 2003 14:37:22 +0300." <3EEEFD72.4040000@intracom.gr> Message-ID: <200306180015.h5I0FIsC083676@possum.icir.org> > XORP installed ok, but when I try to run xorpsh it gives me this: > > step3 (fea) > step3: execute > TaskXrlItem::execute > xrl fea/ifmgr/0.1/commit_transaction?tid:u32=$(interfaces.TID) > +++++++[ 2003/06/17 14:02:08 ERROR fea:19666 FEA +382 > ifconfig_set_ioctl.cc push_config ] On rl0: O/S does not recognise interface > [ 2003/06/17 14:02:08 WARNING fea XrlFeaTarget ] Handling method for > ifmgr/0.1/commit_transaction failed: XrlCmdError 102 Command failed On > rl0: O/S does not recognise interface > TaskXrlItem::execute_done > xrl_done (fea) > TaskManager::task_done > task failed > ############################################################## > ## commit_pass2_done > ## commit failed: 102 Command failed On rl0: O/S does not recognise > interface > MasterConfigTree::config_done: fail: 102 Command failed On rl0: O/S does > not recognise interface > [ 2003/06/17 14:02:08 ERROR rtrmgr:19665 RTRMGR +127 > master_conf_tree.cc config_done ] Startup failed (102 Command failed On > rl0: O/S does not recognise interface) > Startup failed (102 Command failed On rl0: O/S does not recognise interface) > > Obviously I am doing something wrong but, what? > My actual goal is to test MLD on Linux. Are there any shortcuts to that? As Mark said, currently there is no support to start the multicast-related stuff via the router manager. However, there is a shortcut to test MLD only: 1. Get the lastest XORP code from the CVS repository as described in http://www.xorp.org/cvs.html (I just commited some code that fixes an IPv6/MLD-related problem). 2. Apply a simple patch to mld6igmp/test_mld6igmp.cc and mld6igmp/configure_mld6igmp to use MLD instead of IGMP (included at the end of this email). 3. Read mld6igmp/README how to modify/configure mld6igmp/configure_mld6igmp for your needs, and how to start and test MLD. Note that the current shell-script based configuration of MLD is a temporary solution, and the next XORP release should use the router manager instead. Also, note that I have tested MLD only on FreeBSD-4.7 so it may or may not work on IPv6-capable Linux. Please let me know if you run into any problems. Regards, Pavlin Index: configure_mld6igmp =================================================================== RCS file: /usr/local/share/doc/apache/cvs/xorp/mld6igmp/configure_mld6igmp,v retrieving revision 1.11 diff -c -u -r1.11 configure_mld6igmp --- configure_mld6igmp 18 Jun 2003 00:07:03 -0000 1.11 +++ configure_mld6igmp 18 Jun 2003 00:07:41 -0000 @@ -8,12 +8,12 @@ # Send configuration commands to a running MLD/IGMP process. # -MODULE_NAME="IGMP" +MODULE_NAME="MLD" HOSTNAME=`hostname` CLI_TARGET="CLI" -MFEA_TARGET="MFEA_4" -MLD6IGMP_TARGET="IGMP" +MFEA_TARGET="MFEA_6" +MLD6IGMP_TARGET="MLD" . ../cli/xrl_cli_shell_funcs.sh . ../fea/xrl_mfea_shell_funcs.sh Index: test_mld6igmp.cc =================================================================== RCS file: /usr/local/share/doc/apache/cvs/xorp/mld6igmp/test_mld6igmp.cc,v retrieving revision 1.21 diff -c -u -r1.21 test_mld6igmp.cc --- test_mld6igmp.cc 3 Jun 2003 10:30:36 -0000 1.21 +++ test_mld6igmp.cc 18 Jun 2003 00:07:41 -0000 @@ -55,7 +55,7 @@ // // XXX: set to 1 for IPv4, or set to 0 for IPv6 -#define DO_IPV4 1 +#define DO_IPV4 0 // // Local variables From ypol@intracom.gr Wed Jun 18 08:25:22 2003 From: ypol@intracom.gr (Polychronis Ypodimatopoulos) Date: Wed, 18 Jun 2003 10:25:22 +0300 Subject: [Xorp-users] HAVE_IPV6 Message-ID: <3EF013E2.6060203@intracom.gr> Some quick help please? At various places in the code there is this check: #ifdef HAVE_IPV6 ... #endif // HAVE_IPV6 For some reason it doesn't work for me (RH 9.0 with 2.4.20-8) although I do have IPv6 enabled. Is there some starting point where this check is initially made? May I circumvent it? thank you Polychronis From adam@hiddennet.net Wed Jun 18 09:17:01 2003 From: adam@hiddennet.net (Adam Greenhalgh) Date: 18 Jun 2003 09:17:01 +0100 Subject: [Xorp-users] HAVE_IPV6 In-Reply-To: <3EF013E2.6060203@intracom.gr> References: <3EF013E2.6060203@intracom.gr> Message-ID: <1055924221.12278.4.camel@localhost> The configure script detects the presence of ipv6 and writes either /* Define to 1 if you have IPv6 */ /* #undef HAVE_IPV6 */ if you don't have ipv6 or /* Define to 1 if you have IPv6 */ #define HAVE_IPV6 1 if you do. Sorry not upto speed on the other ipv6 bits on linux, but give it a go and see. Adam On Wed, 2003-06-18 at 08:25, Polychronis Ypodimatopoulos wrote: > Some quick help please? > > At various places in the code there is this check: > > #ifdef HAVE_IPV6 > ... > #endif // HAVE_IPV6 > > For some reason it doesn't work for me (RH 9.0 with 2.4.20-8) > although I do have IPv6 enabled. > Is there some starting point where this check is initially made? > May I circumvent it? > > thank you > > Polychronis > > _______________________________________________ > Xorp-users mailing list > Xorp-users@xorp.org > http://mailman.ICSI.Berkeley.EDU/mailman/listinfo/xorp-users From ypol@intracom.gr Wed Jun 18 13:31:59 2003 From: ypol@intracom.gr (Polychronis Ypodimatopoulos) Date: Wed, 18 Jun 2003 15:31:59 +0300 Subject: [Xorp-users] finder Message-ID: <3EF05BBF.1040702@intracom.gr> hello again I am trying to run the configure_mld6igmp script but I run into the following problems: If I run configure_mld6igmp without running rtrmgr first then I get the following: Configuring MLD on pcypol.intranet.gr... cli_add_enable_cli_access_from_subnet4 127.0.0.1/32 [ 2003/06/18 14:35:00 ERROR call_xrl:30244 XRL +246 call_xrl.cc main ] Could not connect to finder. ERROR: failure calling '../libxipc/call_xrl finder://CLI/cli_manager/0.1/add_enable_cli_access_from_subnet4?subnet_addr:ipv4net=127.0.0.1/32' Trying again... Obviously finder is not running. Do I have to bring it up without rtrmgr? Now, if I run rtrmgr first then when I run configure_mld6igmp I get the following: [ 2003/06/18 14:36:35 ERROR call_xrl:30292 XRL +251 call_xrl.cc main ] Connected to finder, but did not become ready. ERROR: failure calling '../libxipc/call_xrl finder://CLI/cli_manager/0.1/add_enable_cli_access_from_subnet4?subnet_addr:ipv4net=127.0.0.1/32' Trying again... So finder is up but still there is some other problem. I tried to have a look into the source code and got up to the FinderClient class. It look like there are no XRLs registered with the finder or something else is wrong. Please help. But since mld6igmp should work without rtrmgr, is there another way to bring the Finder up? Thank you. From lqin@sce.carleton.ca Wed Jun 18 16:09:45 2003 From: lqin@sce.carleton.ca (Liang Qin) Date: Wed, 18 Jun 2003 11:09:45 -0400 Subject: [Xorp-users] a newbie needs help to start Message-ID: <3EF080B9.1050607@sce.carleton.ca> Hi XORP users, I jsut downloaded XORP v0.3 and installed. Now I'd like to run some test cases, could someone tell me where I can find some instructions or document for beginners? TIA! Mike From pavlin@icir.org Wed Jun 18 19:34:34 2003 From: pavlin@icir.org (Pavlin Radoslavov) Date: Wed, 18 Jun 2003 11:34:34 -0700 Subject: [Xorp-users] HAVE_IPV6 In-Reply-To: Message from Polychronis Ypodimatopoulos of "Wed, 18 Jun 2003 10:25:22 +0300." <3EF013E2.6060203@intracom.gr> Message-ID: <200306181834.h5IIYYsC001309@possum.icir.org> > Some quick help please? > > At various places in the code there is this check: > > #ifdef HAVE_IPV6 > ... > #endif // HAVE_IPV6 > > For some reason it doesn't work for me (RH 9.0 with 2.4.20-8) > although I do have IPv6 enabled. > Is there some starting point where this check is initially made? > May I circumvent it? The check for IPv6 existence is performed when ./configure is run. The check itself is in config/acipv6.m4 : the configure script tries to compile and run the following C program: #include #include #include #include #include main() { /* XXX: check if the IPv6 implementation is not missing various pieces. */ int dummy = IP6OPT_MINLEN; dummy = dummy; if (socket(AF_INET6, SOCK_STREAM, 0) < 0) return (1); else return (0); } If the compilation succeeds and the return value from running the program is 0, then it assumes that you have working IPv6, and HAVE_IPV6 is defined in config.h . It could be that the above program is not the apppropriate IPv6 check in case of RedHat 9.0 (e.g, you could have IPv6 without IP6OPT_MINLEN defined). As a quick and less-restricting test, you can compile and run the following program: #include #include #include #include #include #include main() { if (socket(AF_INET6, SOCK_STREAM, 0) < 0) { perror("Failed"); return (1); } else { printf("OK\n"); return (0); } } If it fails, then you indeed do not have IPv6 enabled. If it succeeds, then please let me know and I will fix the original IPv6 test. Thanks, Pavlin From pavlin@icir.org Wed Jun 18 19:40:30 2003 From: pavlin@icir.org (Pavlin Radoslavov) Date: Wed, 18 Jun 2003 11:40:30 -0700 Subject: [Xorp-users] finder In-Reply-To: Message from Polychronis Ypodimatopoulos of "Wed, 18 Jun 2003 15:31:59 +0300." <3EF05BBF.1040702@intracom.gr> Message-ID: <200306181840.h5IIeUsC001383@possum.icir.org> > I am trying to run the configure_mld6igmp script but I run into the > following problems: > > If I run configure_mld6igmp without running rtrmgr first then I get the > following: > > Configuring MLD on pcypol.intranet.gr... > cli_add_enable_cli_access_from_subnet4 127.0.0.1/32 > [ 2003/06/18 14:35:00 ERROR call_xrl:30244 XRL +246 call_xrl.cc main ] > Could not connect to finder. > ERROR: failure calling '../libxipc/call_xrl > finder://CLI/cli_manager/0.1/add_enable_cli_access_from_subnet4?subnet_addr:ipv4net=127.0.0.1/32' > Trying again... > > Obviously finder is not running. Do I have to bring it up without rtrmgr? > > Now, if I run rtrmgr first then when I run configure_mld6igmp I get the > following: > > [ 2003/06/18 14:36:35 ERROR call_xrl:30292 XRL +251 call_xrl.cc main ] > Connected to finder, but did not become ready. > ERROR: failure calling '../libxipc/call_xrl > finder://CLI/cli_manager/0.1/add_enable_cli_access_from_subnet4?subnet_addr:ipv4net=127.0.0.1/32' > Trying again... > > So finder is up but still there is some other problem. > I tried to have a look into the source code and got up to the > FinderClient class. It look like there are no XRLs registered with the > finder or something else is wrong. > Please help. > > But since mld6igmp should work without rtrmgr, is there another way to > bring the Finder up? Did you run first mld6igmp/test_mld6igmp ? For the time being, you need to run this program by hand (i.e., you don't need to run the rtrmgr). This program itself already contains the finder so after you start it you need to run configure_mld6igmp Regards, Pavlin From pavlin@icir.org Wed Jun 18 23:55:36 2003 From: pavlin@icir.org (Pavlin Radoslavov) Date: Wed, 18 Jun 2003 15:55:36 -0700 Subject: [Xorp-users] newbie in XORP In-Reply-To: Message from Polychronis Ypodimatopoulos of "Tue, 17 Jun 2003 14:37:22 +0300." <3EEEFD72.4040000@intracom.gr> Message-ID: <200306182255.h5IMtasC007980@possum.icir.org> > My actual goal is to test MLD on Linux. Are there any shortcuts to that? Polychronis, I have bad news for you. The XORP MFEA IPv6 code is based on Kame's IPv6 multicast kernel (MLD-related definitions, etc), and in case of Linux the names of the relevant constants and headers are either different or are just missing, hence the compilation of the IPv6 multicast-related code will fail on Linux. Further, the Linux kernel doesn't have IPv6 multicast routing support. It appears to have MLD host-side support, but that's all. Currently, the MFEA assumes that the kernel has multicast routing support, and changing it such that it supports MLD only (but no multicast routing) would not be trivial. After all, if someone is running MLD router-side implementation, then it should be within a multicast-capable router. Hence, if you really want to use XORP's MLD router-side implementation, the only thing I can recommend you is to use FreeBSD instead. Pavlin