Timers are validated and never counted: a lab can only show the converged answer #1049

Closed
opened 2026-11-08 08:20:34 +00:00 by jaylouisw · 0 comments
Owner

Every timer in this engine is parsed, range-checked, and then thrown away.

grep -n "forwardDelay\|hello\|dead\|advert" engine/stp.js engine/ospf.js engine/vrrp.js

forward-delay 15, hello 10, dead 40, advertise 1 are all validated — engine/ospf.js:137
even rejects dead < hello — and none of them count for anything, because computeStp() blocks its
ports and computeOspf() installs its routes at t=0.

The consequence is that a lab can only ever show the converged answer. Power on a switch with a
looped topology and it has already converged. Ask for a ping and the network is already up. The
interval every technician actually watches — link light green, nothing passing — cannot be
demonstrated, which is half of what Packet Tracer is for.

What is needed

  • A way to ask a question at an instant: what does this lab do 5 seconds after power-on?
  • The processes that have not finished leave their device in the state the standard puts it in, not
    in a special simulation state: STP ports Listening/Learning and dropping traffic, OSPF without
    learned routes, no DHCP lease, no VRRP mastership.
  • A failure that says converging rather than timed out. Those look identical in a timeline and
    mean opposite things.
  • The default must stay "converged", or every existing lab and test changes behaviour.

Timers that need to be counted

Process Source
STP 802.1D §17.14 — forward delay 15 s, a forwarding port waits two of them
OSPF RFC 2328 hello 10 s, plus the exchange and the SPF hold
VRRP RFC 5798 §6.1 — Master_Down = 3 × Advertisement_Interval + Skew_Time
DHCP the four-message exchange before a lease exists
Every timer in this engine is parsed, range-checked, and then thrown away. ``` grep -n "forwardDelay\|hello\|dead\|advert" engine/stp.js engine/ospf.js engine/vrrp.js ``` `forward-delay 15`, `hello 10`, `dead 40`, `advertise 1` are all validated — `engine/ospf.js:137` even rejects `dead < hello` — and none of them count for anything, because `computeStp()` blocks its ports and `computeOspf()` installs its routes at t=0. The consequence is that a lab can only ever show the converged answer. Power on a switch with a looped topology and it has already converged. Ask for a ping and the network is already up. The interval every technician actually watches — link light green, nothing passing — cannot be demonstrated, which is half of what Packet Tracer is for. ## What is needed - A way to ask a question at an instant: what does this lab do 5 seconds after power-on? - The processes that have not finished leave their device in the state the standard puts it in, not in a special simulation state: STP ports Listening/Learning and dropping traffic, OSPF without learned routes, no DHCP lease, no VRRP mastership. - A failure that says *converging* rather than *timed out*. Those look identical in a timeline and mean opposite things. - The default must stay "converged", or every existing lab and test changes behaviour. ## Timers that need to be counted | Process | Source | |---|---| | STP | 802.1D §17.14 — forward delay 15 s, a forwarding port waits two of them | | OSPF | RFC 2328 hello 10 s, plus the exchange and the SPF hold | | VRRP | RFC 5798 §6.1 — Master_Down = 3 × Advertisement_Interval + Skew_Time | | DHCP | the four-message exchange before a lease exists |
Sign in to join this conversation.
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
jaylouisw/packetforge#1049
No description provided.