Missing backport for scutum and circinus https://github.com/vyos/vyos-build/commit/bd0de0defd2ba7c7753495e0f831805c4b6de89a
- Feed Queries
- All Stories
- Search
- Feed Search
- Transactions
- Transaction Logs
Wed, Nov 18
Tue, Nov 17
Patch
Mon, Nov 16
This can be reproduced using the following example certificate:
Tested on a 1.5.1-S1 router that is forwarding traffic. Manually patched the CHUNK_SIZE, but the patch applied to progressbar.py fine.
Not a thorough test, but I tried it on a 1.5.0-S1 Proxmox VM. The patch didn't change the chunk size, but it did patch the progressbar.py file.
@rherold I tried the accel-ppp side you described, on the two machines I have, and it works.
Ran accel-ppp-ng PR #42 at 7e24429 on VyOS rolling 2026.11.16 x86 with its VPP 25.10 package: 10,000 sessions at 200/s with 30 s interims came up in 56 s, 64,000 at 1000/s in 140 s, no session with a 0 counter, every Interim-Update carried traffic, and no session was ended by a counter read. The rolling package's own accel-ppp-ng 1.13.0 ends every session with an interim interval at the first interim (Lost-Carrier, 0 octets). Details: https://github.com/accel-ppp/accel-ppp-ng/pull/42#issuecomment-6908251592
Sun, Nov 15
In T9621#286826, @Viacheslav wrote:Did it worked in any version or should we have some verify strep to avoid commits with !xxx ?
@Viacheslav Thanks, you are right, and I checked it without VPP. On a 2026.11 rolling image (two PPPoE clients, one on eth1 and one on eth2, kernel PPPoE, no vpp configuration) I removed one interface with delete service pppoe-server interface eth2 and committed:
Did it worked in any version or should we have some verify strep to avoid commits with !xxx ?
I expect this behaviour even without VPP
accel-ppp-ng probably cannot reread interface configuration without restarting
Sat, Nov 14
Fri, Nov 13
FTR, the bug fix is upstream 8aecf0bbcc72
I managed to reproduce this issue and pin-point its root cause.
Actually this bug is already fixed upstream since 6.6.160.
I tested in VyOS 1.5.1-S1 and got the same error.
Suggesgted patch to test
Thu, Nov 12
I think this might be related to some problems I’ve noticed as well.