Cisco – MTU between Juniper and Cisco, cause of packet loss

You stated that ports 3 and 42 were configured on the Catalyst switch, but then provided configurations for ports 46 and 48. The configuration you posted for port 46 should be applied to port 3 that connects to the EX2200. Your router's connection is unchanged, so hopefully we can assume that configuration is fine.

Now, on the EX2200, the following lines of code would be appropriate to do the following:

ge-0/0/0 - trunk allowing the same vlans as defined above on port 46

ge-0/0/6 - access port on VLAN80

set vlans vlan80 vlan-id 80
set vlans vlan82 vlan-id 82
set vlans vlan83 vlan-id 83
set vlans vlan93 vlan-id 93
set vlans vlan289 vlan-id 289
set interfaces ge-0/0/0 description uplink-to-catalyst 
set interfaces ge-0/0/0 unit 0 family ethernet-switching port-mode trunk
set interfaces ge-0/0/0 unit 0 family ethernet-switching vlan members vlan80
set interfaces ge-0/0/0 unit 0 family ethernet-switching vlan members vlan82
set interfaces ge-0/0/0 unit 0 family ethernet-switching vlan members vlan83
set interfaces ge-0/0/0 unit 0 family ethernet-switching vlan members vlan93
set interfaces ge-0/0/0 unit 0 family ethernet-switching vlan members vlan289
set interfaces ge-0/0/6 unit 0 family ethernet-switching port-mode access
set interfaces ge-0/0/6 unit 0 family ethernet-switching vlan members vlan80

Some other suggestions for you:

1) Turn on LLDP on your switch so you can do a show lldp neighbors and see where your connections go.

2) Don't use RSTP for spanning tree on the juniper switch, it doesn't play nice with Cisco that well, use VSTP instead. If you end up with a ton of vlans, you might even need to use MSTP.

3) Turn off chassis alarm for the management ethernet if you're not using it.

On the EX2200:

delete protocols rstp
set protocols vstp vlan all bridge-priority 4k
set protocols lldp interface all
set chassis alarm management-ethernet link-down ignore

On the Catalyst (if it supports it)

lldp run

Any idea on resolving these collisions?

The simple answer, remove the manual speed / duplex configurations from the interfaces affected by the collisions. Auto negotiation is required by Gigabit Ethernet in the IEEE spec.

Almost all ethernet interfaces default to autonegotiation. If you have configured manual speed / duplex on your router interfaces, then autonegotiation on the other side will fail (usually falling back to half duplex). The mismatched duplex results in collisions (and thus packet loss)...

There are very few reasons to hard-code speed / duplex these days. In the early days of autoneg, it was not well-standardized and did not work well; however, those days are long gone. Use auto-negotiation everywhere... and tell your friends.