SyncLackResc problems on BN router

[Garrie]

I need some help with a throughput problem I am having on a BN router running ver 13.20.
The router has 8 slots in use with a couple of 10M Quad Ethernet cards and the rest Qsync. Most of the sync ports are running low speed X.25 with three running Direct Access Frame Relay with 3 or 4 pvcs and supporting IP,OSPF and IPX. All processor cards are FREII-040 with 16MB mem. I'm pretty sure the current set-up should not be overworking the router.
I'm having loads of trouble with the FR ports. The same is happening on each F/R port but I'll concentrate on one which is by far the worst - (happens to be the most utilised one). It has a 2Mb access speed and 3 pvcs configured for direct access of 1024Mb (cir 512k), 384k (cir192k) and 256k (cir 128k). Total pvc throuput with traffic shaping should be 1664k. What I'm seeing is loads of Sync Lack Resouce errors on both Tx and Rx when the utilisation gets up towards the max. I would have thought that the sync interface could easily handle this. I have traffic shaping on for all pvcs, I am also using WCP and ProtPri. The Frame Relay network is not overloaded and I am not getting any FECN's/BECN's.
I have increased the SyncTxQueue to 63 and have carved the memory to increase the Global to 8Mb. Also played around with all traffic shaping and WCP off/on and anything else I could think of but to no avail.

The dropped packets are starting to cause the client problems and it is acknowledged that more bandwidth is required on the 1024Mb pvc so we are adding a new FR circuit of 2Mb to replace the 1024Mb pvc. My concern is that if the current interface cannot handle 1.6 MB throughput am I going to see the same thing on the new interface trying to push 2Mb?
Any help in overcoming these SyncLackResc problems would be greatly appreciated. Do I need more mem on the cards?

 

[Googer]

I experienced something simular on a HSSI port using Frame Relay. My write up on it was as follows and how it was fixed.

Good Luck.

Problem and Solution:

The problem is that site manager does not work properly to set the queue levels properly on this interface. You have 1200 queues to use on a HSSI card on a BN router, however it does not matter how I set them they were still using the default values. Remember, with frame relay the number of queues you set for the interface is divided by the number of PVCs on that interface (EX: You set high queue to 300 and you have 30 PVCs, you will 10 queues per PVC). You can check the vaulues with the folling MiB get. This appears to only be a problem on Frame Relay interfaces using Protocol Priority.

get wfFrIfDlcmiEntry.*.710101.0 (710101.0 is the interface number, this will vary by interface and router)

You want to look at these three fields to confirm their sizes.

wfFrIfDlcmiEntry.wfFrDlcmiShapingHiQueueLimit.710101.0 = 30
wfFrIfDlcmiEntry.wfFrDlcmiShapingNormalQueueLimit.710101.0 = 200
wfFrIfDlcmiEntry.wfFrDlcmiShapingLoQueueLimit.710101.0 = 30

To set the values use the folowing commands.

set wfFrIfDlcmiEntry.39 (high)
set wfFrIfDlcmiEntry.40 (Normal)
set wfFrIfDlcmiEntry.41 (low)

NOTE: You need to add the interface number to this set command.


This will set queues to the proper values. Make sure to confirm this by doing another get wfFrIfDlcmiEntry.*.(interface number)


Useful commands:
show fr stats shaping high
get wfCctOptsEntry.*.2 6 (6 is number of interface)
list -i wfFrVCircuitEntry (show list of interfaces)
get wfFrVCircuitEntry.*."Interface Number" (show VC interface stats)