Open jhlasnik opened 2 years ago
I think that kernel version got updated but they didn't bump the version number, so I didn't detect it. I've build the new kernel image from the kernel extracted from 5.0 image: networkop/cl-kernel:5.0.1
Can you try specifying kernel: docker.io/networkop/cl-kernel:5.0.1
and see if you still see any errors?
btw, I've discovered another bug that manifests itself when you don't set the system-mac explicitly. If you're seeing an error when ifreload tries to set a system mac, try using networkop/cx:5.0.4
Yeah I saw a similar issue when I had it set, I was trying a bunch of different combinations for this Inter-op lab with Arista.
I’ll try it out tonight and report back! Thanks!
Best regards, Joe Hlasnik
From: Michael Kashin @.> Sent: Tuesday, June 21, 2022 12:11:07 PM To: networkop/cx @.> Cc: Hlasnik, Joseph @.>; Author @.> Subject: Re: [networkop/cx] Netlink Errors for SVIs (Issue #7)
[ CAUTION: External Email ]
btw, I've discovered another bug that manifests itself when you don't set the system-mac explicitly. If you're seeing an error when ifreload tries to set a system mac, try using networkop/cx:5.0.4
— Reply to this email directly, view it on GitHubhttps://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnetworkop%2Fcx%2Fissues%2F7%23issuecomment-1161964775&data=05%7C01%7Cjoseph.hlasnik%40expedient.com%7C51ee2775c545479e80d008da53a0a6f3%7Cdaaf576f055447d487baf848642811e0%7C0%7C0%7C637914246713581570%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=jx77mP5TxQ1BHfjDZYnqrZEODNLXwRVDUfXbFD2z2gg%3D&reserved=0, or unsubscribehttps://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FAF5NS4PHOD32YGAT6SGURGDVQHSRXANCNFSM5ZKBVGXQ&data=05%7C01%7Cjoseph.hlasnik%40expedient.com%7C51ee2775c545479e80d008da53a0a6f3%7Cdaaf576f055447d487baf848642811e0%7C0%7C0%7C637914246713581570%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=E7%2Bmi%2FegAUAMXALXMIhFIFySLQrWvqI2z6OcTcFNcPg%3D&reserved=0. You are receiving this because you authored the thread.Message ID: @.***>
So I did test setting the kernel on 5.0.1 and 5.0.4 and I'm still seeing those same errors on bootup with the system mac set.
root@clab-cx-sw1:mgmt:~# more /var/log/syslog | grep -E NetlinkError
2022-06-22T13:20:55.331706+00:00 1d58c2642a4b6fbd networking[4057]: raise NetlinkError(e, "cannot set dev %s address %s" % (ifname, hw_address), ifname=ifname)
2022-06-22T13:20:55.331849+00:00 1d58c2642a4b6fbd networking[4057]: lib.nlcache.NetlinkError: netlink: br_default: cannot set dev br_default address 89:b8:73:18:ad:e3: operation failed with 'Cannot assign requested address' (99)
2022-06-22T13:20:55.497630+00:00 1d58c2642a4b6fbd networking[4057]: raise NetlinkError(e, "cannot set dev %s address %s" % (ifname, hw_address), ifname=ifname)
2022-06-22T13:20:55.497739+00:00 1d58c2642a4b6fbd networking[4057]: lib.nlcache.NetlinkError: netlink: vlan10: cannot set dev vlan10 address 89:b8:73:18:ad:e3: operation failed with 'Cannot assign requested address' (99)
2022-06-22T13:20:55.636432+00:00 1d58c2642a4b6fbd networking[4057]: raise NetlinkError(e, "cannot set dev %s address %s" % (ifname, hw_address), ifname=ifname)
2022-06-22T13:20:55.636611+00:00 1d58c2642a4b6fbd networking[4057]: lib.nlcache.NetlinkError: netlink: vlan30: cannot set dev vlan30 address 89:b8:73:18:ad:e3: operation failed with 'Cannot assign requested address' (99)
2022-06-22T13:20:55.882610+00:00 1d58c2642a4b6fbd networking[4057]: raise NetlinkError(e, "cannot set dev %s address %s" % (ifname, hw_address), ifname=ifname)
2022-06-22T13:20:55.882778+00:00 1d58c2642a4b6fbd networking[4057]: lib.nlcache.NetlinkError: netlink: br_l3vni: cannot set dev br_l3vni address 89:b8:73:18:ad:e3: operation failed with 'Cannot assign requested address' (99)
2022-06-22T13:20:56.126300+00:00 1d58c2642a4b6fbd networking[4057]: raise NetlinkError(e, "cannot set dev %s address %s" % (ifname, hw_address), ifname=ifname)
2022-06-22T13:20:56.126508+00:00 1d58c2642a4b6fbd networking[4057]: lib.nlcache.NetlinkError: netlink: vlan20: cannot set dev vlan20 address 89:b8:73:18:ad:e3: operation failed with 'Cannot assign requested address' (99)
2022-06-22T13:20:56.278890+00:00 1d58c2642a4b6fbd networking[4057]: raise NetlinkError(e, "cannot set dev %s address %s" % (ifname, hw_address), ifname=ifname)
2022-06-22T13:20:56.279057+00:00 1d58c2642a4b6fbd networking[4057]: lib.nlcache.NetlinkError: netlink: vlan40: cannot set dev vlan40 address 89:b8:73:18:ad:e3: operation failed with 'Cannot assign requested address' (99)
# topology documentation:
name: cx
topology:
nodes:
sw1:
kind: cvx
image: networkop/cx:5.0.4
kernel: docker.io/networkop/cl-kernel:5.0.1
binds:
- hosts/clab-cx-sw1/startup.yaml:/etc/nvue.d/startup.yaml
sw2:
kind: cvx
image: networkop/cx:5.0.4
kernel: docker.io/networkop/cl-kernel:5.0.1
binds:
- hosts/clab-cx-sw2/startup.yaml:/etc/nvue.d/startup.yaml
can you share the SVG file you get using systemd-analyze plot > boot.svg
?
File attached
it seems to work fine in my case. I think it could be down to the background CPU noise and race conditions.
It will work if you manually do a systemctl restart sys-mac-shot
. As you can see from this graph, I've tried to put it as close to the end as possible.
The right way for me would be to try and locate how/why the wrong(multicast) MAC is being set in /run/system_mac
and try to correct/patch that. However so far I couldn't find it.
So what I'll do instead is release another version where I'll delay the sys-mac-shot
execution by, say 10 seconds. This happens after network-online
so it should delay the time to login.
networkop/cx:5.0.5
is out with the patch.
Looks like I've found where the MAC bug is coming from. https://github.com/networkop/cx/blob/5f8dc609a3a0c7f5d16c6cfc935100f49332ab09/hacks/decode-syseeprom.5.1#L373 and https://docs.python.org/3/library/uuid.html#uuid.getnode
If all attempts to obtain the hardware address fail, we choose a random 48-bit number with the multicast bit (least significant bit of the first octet) set to 1 as recommended in [RFC 4122](https://tools.ietf.org/html/rfc4122.html).
thanks to https://stackoverflow.com/a/159195
this should be fixed in networkop/cx:5.0.6
Hello,
I'd like to report an issue I'm seeing on the 5.0.x images. I'm still working my way through learning the config so this certainly could be me missing something but wanted to report it as we discussed.
Full config and syslog files attached.
If you need any more information please let me know! syslog.log config.txt