11-20-2012 07:29 AM
On one of our Brocade (switchType: 83.3) we have a GE port dedicated for FCIP replication for our IBM storwize (100km distant).
On this port we get regular GE FIFO overflow errors:
> portstatsshow ge0
ge_stat_tx_frms 163267708285 GE transmitted frames
ge_stat_tx_octets 168983842881620 GE transmitted octets
ge_stat_tx_ucast_frms 163267485251 GE transmitted unicast frames
ge_stat_tx_mcast_frms 0 GE transmitted multicast frames
ge_stat_tx_bcast_frms 115734 GE transmitted broadcast frames
ge_stat_tx_vlan_frms 0 GE transmitted vlan frames
ge_stat_tx_pause_frms 107300 GE transmitted pause frames
ge_stat_rx_frms 104557403739 GE received frames
ge_stat_rx_octets 21666295301492 GE received octets
ge_stat_rx_ucast_frms 104212900056 GE received unicast frames
ge_stat_rx_mcast_frms 162829338 GE received multicast frames
ge_stat_rx_bcast_frms 163326966 GE received broadcast frames
ge_stat_rx_vlan_frms 0 GE received vlan frames
ge_stat_rx_pause_frms 0 GE received pause frames
ge_err_carrier 0 GE lost carrier sense
ge_err_length 0 GE invalid length
ge_err_crc 0 GE CRC Errors
ge_err_abort 0 GE abort frames
ge_err_overrun 0 GE overruns
ge_err_fifo_ovf 18350364 GE Fifo overflow
I just can't find anything on internet that could explain it.
That GE port is used for FCIP tunneling of this VE port:
> portcfgshow 16
Area Number: 16
Speed Level: AUTO(HW)
Fill Word: 0(Idle-Idle)
AL_PA Offset 13: OFF
Trunk Port ON
Long Distance OFF
VC Link Init OFF
Locked L_Port OFF
Locked G_Port OFF
Disabled E_Port OFF
Locked E_Port OFF
ISL R_RDY Mode OFF
RSCN Suppressed OFF
Persistent Disable OFF
LOS TOV enable OFF
NPIV capability ON
QOS E_Port OFF
Port Auto Disable: OFF
EX Port OFF
F_Port Buffers OFF
Fault Delay: 0(R_A_TOV)
NPIV PP Limit: 126
Does anybody has any idea of where these FIFO overflow errors could come from? Could that be due to a bad setting of the BB credit on the VE port?
Thanks for your help and don't hesitate if more logs are needed.
11-20-2012 01:25 PM
--->>> Does anybody has any idea of where these FIFO overflow errors could come from?
I think,.... but I'm not 100% sure, those Fifo owerflow are caused/coming from a invalid lenght, caused by to many Frame sending.
11-21-2012 10:17 AM
Thanks for your help. Digging a bit we're wondering if that could come from a bad MTU parameter on our tunnel.
Right now it is set at 1500 and I read that it might be worse updating it to a bit less than 2500 (provided that all the devices in the network supports jumbo frame). Is there a way of updating this parameter without recreating the entire tunnel?
11-21-2012 10:26 AM
--->>>Digging a bit we're wondering if that could come from a bad MTU parameter on our tunnel.
Is most probable, i facing time ago by customer same experience and the problem was wrong set MTU value.
In my case the Carrier instist first, the WAN Router was configured per default with 1500.
we ask again and again, and later then confirm the MTU was set with another value.
Jumbo Frame shold be set enable.
11-26-2012 08:18 AM
Thanks for your feedback.
Unfortunately, our switch is a 7800 (type 83.3) so the MTU can't be set to more than 1500...
So, I'm back to my first suggestion : changing the BB Credit value on the VE port (16) that is linked to this ge port for FCIP. It is set to 8 on this 1Gb link to a site 100km away so my guess is 50 would fit better.
a) does BB credit is used on a VE port? (the GUI states that it is for F-port)
b) would changing the BB credit reset the tunnel?
c) would activating compression solve the issue?
Today, the line went so overloaded due to a DRP testing that it went down for a few seconds...
For additional information, I have attached a screenshot showing that our Storwize which is directly connected to this 7800 on ports 0-3 has some BB credit Zero timer going really high (up to 160 seconds). I take it as back pressure from the issues described precedently.
We're already planning a time slot to change the BB credit value so I'll keep you posted but I wanted to share my thoughts and the status in case someone else faces these errors.
12-03-2012 03:01 AM
We have done the descibed changes this week-end :
- BB credit can not be changed / activated on a VE port;
- compression has been activated (moderate);
No improvement, still lots of FIFO overflow errors, Out Of Order packets, Zero BB credit timer at 190seconds... and replication group falling...
No idea what's going on so we might open a case at Brocade...