-- Evgeniy Polaykov ( s0mbre ) Crash is better than data corruption. -- Art Grabowski Attachment: carp.c Description: Text Data Attachment: carp.h Description: Text Data Attachment: carp_ioctl.h Desc
Why do you need to put this stuff in the kernel? This should be implemented just the same way as VRRP was - in user space. BTW, is there a spec for this protocol or its one of those things where you
Hmm... Just because i think it works better being implemented in the kernel? :) I don't think it is a good answer thought. It is faster, it is more flexible, it has access to kernel space... Exactly
Just an addition[from private e-mail]: UCARP Userspace can't in principle. Current kernel implementation can't too, but it can. In principle. But better implementation should use both carp and ct_syn
Yeah, I know ;-> and probably thats what the opnebsd people did. I still think it should live in user space. This should apply to anything thats control related because such things tend to be contino
Load balancing between different computers? How nodes will know about each other using only tc extension? Kernel traps packet, send info about it to userspace, it decides drop it or not... Not very f
Userspace is too slow. It can only initiate master's failover, load balancing is a good example here - userspace _itself_ can not control real time traffic. ct_sync module does this. It uses connecti
What is it that CARP does that couldnt be achieved by VRRP? VRRP is implemented in user space. As a user myself i can assure you even with heartbeats at 100ms granularity i have seen no issues which
Why do they need to know about each other. Maybe explain a little how said load balancing is achieved. I am hoping CARP knows how to deal with dropped packets. use socket to send to user space. When
I will answer a question by question, sorry. Has vrrp some authentification mechanism? Can it be used over IPv6? (CARP also can't but it is _very_ easy to add, I just don't have IPv6 network setup to
Kind of scuch scenario: If I am a masater, than get half of bandwidth, but if slave count is less than threshold than get more. If I am a slave and slave count is more than threshold than get 0.5/sla
Ok, so some controller is in charge - seems like thats something that could be easily done in user space based on mastership transitions. Ok, i wont tell ;-> In VRRP for example its the number of hea
They (at least used to) claim to be able to do so. Theres effort to have it do v6. http://www.ietf.org/internet-drafts/draft-ietf-vrrp-ipv6-spec-06.txt I agree its lame to have it as an after thought
Yes, but here is tricky but true example: Some time ago e1000 driver from Intel had possibility to do hardware bonding(i absolutely don't remember how it was called, but idea was the same as in bondi
I remeber that cruft. Actually (sadly) people like montavista ship that thing in their distros (under the disguise of carrier grade linux;->). I think the current folks out of intel working on Linux
<arguments and contrarguments are skipped> I saw it... May I not comment it? I do not want to look like rude freak... :) Only for now, since we can imagine only some examples now. When number of agen
[..] Ok, you are right. I do think that there are people who have run this over IPSEC though. I could swear that the current linux based one does. I wish we could get Alexander to comment on this dis
for something? MICROKERNEL, I see it :) Non broacast/multicast will _strongly_ complicate protocol. Broadcast will waste apprication/kernel "bandwidth". Is it is. CARP may do it, but it requires in-k