How do I fix Eobc errors?

[mctiesman]

Recently received a problem ticket indicating a threshold violation on Eobc 0/0. This is occurring on a 6509 with software 12.2(33)SXI8. When doing a 'show eobc" I see the following:

Interface information:

Interface EOBC0/0 (idb = 0x50AD5524)

Hardware is Mistral EOBC (revision 5)

Address is 0000.1500.0000 (bia 0000.1500.0000)

Encap size         = 14         hardware status  = 0x210840

IDB type           = 18         IDB state        = 4

Encap type         = 0x1        Span encap size  = 0

Error threshold    = 5000       Error count      = 0

Counters:

rxring             = 0x921D940  rx ring entries       = 512       

rx_head            = 248        rx_tail               = 0

inputs             = 7041590557 rx_cumbytes           = 2685550134617

hw inputs          = 0          hw rx_cumbytes        = 0

rx rate (bits/sec) = 629000     rx rate (packets/sec) = 216

rx_buf_unavail     = 0          rx input drops        = 299       

input broadcast    = 25         input resource        = 1531956409

input error        = 0          input giants          = 0

input crc          = 299        rx illegal length     = 0

rxr eobc shadow    = 0x50BE2C20 txr eobc shadow       = 0x47B3DC00

txring             = 0x921F980  tx ring entries       = 0x200

tx_head            = 52         tx_tail               = 52

outputs            = 6628559346 tx_cumbytes           = 559065790427

hw outputs         = 0          hw tx_cumbytes        = 0

tx rate (bits/sec) = 130000     tx rate (packets/sec) = 207

tx_retry_error     = 471        tx_retry_count        = 17642354

tx_process_stopped = 96         tx total drops        = 0

Mistral Registers

soft_reset_cfg     = 0x040000   dma_buffer_size_reg   = 0x000000

int_mask_hi        = 0x00007E   int_mask_lo           = 0xE7001A58

rxdscp_cnt         = 512        txdscp_cnt            = 0

rxwork_dscp        = 0xE8C0     txwork_dscp           = 0xFB20

mistral_eobc_ds    = 0x47A30278 mistral_dma_register  = 0x30000000

mistral_glbl_reg   = 0x10020000

Misc. Global Registers:

global_cfg         = 0x20       mis_init_sts          = 0xF

        dimm_parm_cfg_hi   = 0x00000576 dimm_parm_cfg_lo      = 0x42040F5A

tm_init_size_cfg   = 0x8000

The counters are not registering at this time. What does this mean and how to I resolve this? Can you clear the eobc counters? Thanks!!

[educruz]

Hey!

As we know, EOBC is a bus-based Ethernet. In a shared-medium bus like EOBC, we expect to see some type of errors, such as CRC errors, sometimes.  As long as the number of errors does not increase very quickly in a short period of time, we do not need to be concerned. There is no impact on the operation of the router when we observe a CRC errors because the protocols on the EOBC have a retry mechanism.

When several keep-alives are missed, it means communication is lost between the Application-Specific Integrated Circuits (ASICs) and the supervisor. In order to prevent this, the code contains a workaround that resets the EOBC when it sees that the messages do not increment constantly.

EOBC is a half-duplex out of band channel which is used for backplane communication between SP/RP and line cards. These errors are mostly benign, as deferred implies packet is ready but the media to send it on is busy. The reason for this is that the EOBC is a half-duplex interface, and the fact that there are deferred simply means that we are receiving when we attempt to transmit. With a half-duplex interface we cannot do that, so we defer the packet, and wait. Likewise, on a half-duplex interface collisions happen, and are part of normal operation.

Kind regards,

- Ed.