Hi All, I'm having trouble getting RHEL to re-instate persistent IPV6 routes on reboot. I have the routes set up in /etc/sysconfig/network-scripts/route6-eth0 but they aren't getting applied on startup. But when I do an ifdown followed by an ifup, or a 'service network restart' I get the following message: RTNETLINK answers: No route to host ..Which is fair enough, as it seems that the IPV6 network isn't actually ready when the route is added. Has anybody got this working before? Is there somewhere else I should be adding the routes? Cheers, Leigh.
Leigh Sharpe <leigh.sharpe@gmail.com> wrote:
I'm having trouble getting RHEL to re-instate persistent IPV6 routes on reboot. I have the routes set up in /etc/sysconfig/network-scripts/route6-eth0 but they aren't getting applied on startup. But when I do an ifdown followed by an ifup, or a 'service network restart' I get the following message:
RTNETLINK answers: No route to host
..Which is fair enough, as it seems that the IPV6 network isn't actually ready when the route is added.
Could your script sleep for a few seconds until the network is up?
It's not actually my script. It's ifup-ipv6 in /etc/sysconfig/network-scripts. I'm thinking at this point I should probably put an 'ip -6 route add' in /etc/rc.local, but it just doesn't feel right. On Fri, Feb 10, 2012 at 12:10 PM, Jason White <jason@jasonjgw.net> wrote:
Leigh Sharpe <leigh.sharpe@gmail.com> wrote:
I'm having trouble getting RHEL to re-instate persistent IPV6 routes on reboot. I have the routes set up in /etc/sysconfig/network-scripts/route6-eth0 but they aren't getting applied on startup. But when I do an ifdown followed by an ifup, or a 'service network restart' I get the following message:
RTNETLINK answers: No route to host
..Which is fair enough, as it seems that the IPV6 network isn't actually ready when the route is added.
Could your script sleep for a few seconds until the network is up?
_______________________________________________ luv-main mailing list luv-main@luv.asn.au http://lists.luv.asn.au/listinfo/luv-main
Leigh Sharpe <leigh.sharpe@gmail.com> wrote:
It's not actually my script. It's ifup-ipv6 in /etc/sysconfig/network-scripts.
I'm not familiar with the RHEL file system layout or with what belongs to packages. I would at least suggest lodging a bug report against the script, searching Red Hat's bug database first in case it's already done.
On Friday 10 February 2012 12:49:25 Jason White wrote:
I would at least suggest lodging a bug report against the script, searching Red Hat's bug database first in case it's already done.
Remember that public bugzilla reports are generally not really looked at for RHEL, you need to lodge a support request instead. Not that support requests seem to get treated much better, November 2010 we reported a bad kernel bug in 5.5 (packets on a dual port Mellanox 10gigE adapter get delivered to the wrong interface, almost at random) which has just got push back yet again, this time to RHEL 5.9, as it requires RH to update to a Mellanox driver that was already publicly available when the bug was reported. :-( cheers, Chris -- Chris Samuel : http://www.csamuel.org/ : Melbourne, VIC This email may come with a PGP signature as a file. Do not panic. For more info see: http://en.wikipedia.org/wiki/OpenPGP
On Fri, 10 Feb 2012, Chris Samuel wrote:
On Friday 10 February 2012 12:49:25 Jason White wrote:
I would at least suggest lodging a bug report against the script, searching Red Hat's bug database first in case it's already done.
Remember that public bugzilla reports are generally not really looked at for RHEL, you need to lodge a support request instead.
Not that support requests seem to get treated much better, November 2010 we reported a bad kernel bug in 5.5 (packets on a dual port Mellanox 10gigE adapter get delivered to the wrong interface, almost at random) which has just got push back yet again, this time to RHEL 5.9, as it requires RH to update to a Mellanox driver that was already publicly available when the bug was reported. :-(
I personally don't understand why managers insist on using RHEL. Sure, I understand that they want to cover their arses. No one ever got fired for buying RHEL, as they say. But no one I have ever come across has ever gotten a bug fix included in RHEL after submitting a support case (other than a relatively minor bug 463880, which only took 3 years from when I reported it with pointer to the appropriate patch, before it finally got pushed to servers that we could pull from). So that shiny expensive arse covering you bought turned out to be completely useless! Meanwhile, those lucky folks whose bosses allow them to use debian in their clusters get to enjoy most bugs being fixed within a few months for the next dot release of stable if the bug is important enough. Sure, the hardware venduhs will say it's not officially supported, but it works in practice and there's a strong community presense on the community mailing list on the venduh pages. All without having to involve lawyers and beancounter and tender processes to draft up a support contract. -- Tim Connors
participants (4)
-
Chris Samuel -
Jason White -
Leigh Sharpe -
Tim Connors