Re: [Freeipmi-users] libipmiconsole: Duplicate data received by target after NACK

[Albert Chu]

Hi Dave,

It's hard to say what your problem is given the limited information.  I
assume that SOL w/ something like the 'ipmiconsole' tool works just fine
w/ this system?  It's only when scripting it things go bad?

Underneath the covers libipmiconsole will resend data if the data is
lost or not ACKed properly, so that may explain some of your duplicate
data in your script.

If by chance things are timing out too quickly on your network, you
could play with a number of the timeout settings in libipmiconsole.
Perhaps things are being sent/resent too aggressively for your
environment.

Al

On Mon, 2014-09-08 at 10:21 +1000, David O'Shea wrote:
> Hi all,
> 
> I'm hitting another problem with my Python script using IPMI SOL to
> talk to a Dell system via its iDRAC.  Once the Dell system is running
> Linux, I'm trying to write a small script to it.  Initially I had lots
> of problems where the Dell system never received any of the data I
> sent, and I'm not sure how flow control works for SOL (I tried running
> 'stty crtscts' on the Dell system but it didn't help), so I started
> pacing the output by sending only 8 bytes every 100ms, which should be
> fine at 115200bps.
> 
> One time this error appeared whilst sending this small script:
> 
> (ipmiconsole_processing.c, _sol_bmc_to_remote_console_packet, 2672): 
> hostname=silicon-bmc.adl.quantum.com; protocol_state=9h: SOL packet NACKED: 
> accepted_character_count: accepted_character_count = 0,  
> c->session.sol_input_character_data_len = 8
> 
> The Dell system then appeared to receive one of these 8-byte chunks
> twice, presumably the one that was NACKED.
> 
> First, here are two 8-byte chunks that were echoed back normally:
> 
> [2014-09-07 22:15:27,095] DEBUG - Send chunk to target: 'on" in\n '
> [2014-09-07 22:15:27,129] DEBUG - Raw input from target: 'on" in\r\n '
> [2014-09-07 22:15:27,195] DEBUG - Send chunk to target: '   PREIN'
> [2014-09-07 22:15:27,229] DEBUG - Raw input from target: '   PREIN'
> 
> so it looks like I normally get a response in 35ms, but then this happened:
> 
> [2014-09-07 22:15:27,296] DEBUG - Send chunk to target: 'IT)\n    '
> [2014-09-07 22:15:27,396] DEBUG - Send chunk to target: '    # Th'
> [2014-09-07 22:15:27,497] DEBUG - Send chunk to target: 'is is us'
> [2014-09-07 22:15:27,597] DEBUG - Send chunk to target: 'ed to pr'
> [2014-09-07 22:15:27,698] DEBUG - Send chunk to target: 'epare th'
> [2014-09-07 22:15:27,798] DEBUG - Send chunk to target: 'e interf'
> [2014-09-07 22:15:27,819] DEBUG - Raw input from target: 'IT)\r\n    '
> [2014-09-07 22:15:27,869] DEBUG - Raw input from target: 'IT)\r\n        # 
> This is used to prepare the interf'
> 
> No responses received for over 500ms, then one of the 8-byte chunks is echoed 
> back twice.
> 
> I verified that "IT)\r\n    " was written to the file on the target twice.
> 
> There is no retry mechanism in my script so I assume that libipmiconsole is 
> resending when it receives a NACK.
> 
> Do I need a full verbose log in order to debug this properly?
> 
> I went back through my logs and found another instance of one of these 
> "NACKED" messages, it was for 31 characters but those characters didn't seem 
> to get duplicated.
> 
> Thanks in advance,
> David
>                                         
> _______________________________________________
> Freeipmi-users mailing list
> address@hidden
> https://lists.gnu.org/mailman/listinfo/freeipmi-users
-- 
Albert Chu
address@hidden
Computer Scientist
High Performance Systems Division
Lawrence Livermore National Laboratory