#114 1.8.12 sol very slow

version-1.8.13
closed-fixed
None
5
2013-07-06
2012-10-10
arekm
No

ipmitool sol is insanely slow, printing characters or part of lines once every few seconds, most of serial console lines are getting missed.

There is no such problem with ipmitool 1.8.11 sol, so looks like a regression.

I'm connecting to S5000VSA :

Device ID : 32
Device Revision : 1
Firmware Revision : 0.64
IPMI Version : 2.0
Manufacturer ID : 343
Manufacturer Name : Intel Corporation
Product ID : 43 (0x002b)
Product Name : Unknown (0x2B)
Device Available : yes
Provides Device SDRs : no
Additional Device Support :
Sensor Device
SDR Repository Device
SEL Device
FRU Inventory Device
IPMB Event Receiver
Chassis Device
Aux Firmware Rev Info :
0x0b
0x41
0x00
0x03

Discussion

  • arekm

    arekm - 2012-10-10

    Reverting commit below makes things work well again:

    http://ipmitool.cvs.sourceforge.net/viewvc/ipmitool/ipmitool/lib/ipmi_main.c?r1=1.25&r2=1.26

    Author: blaism <blaism>
    Date: Tue Apr 27 16:53:05 2010 +0000

    Changed default cipher suite to 1 instead of 3 for iol20

    diff --git a/lib/ipmi_main.c b/lib/ipmi_main.c
    index 3d93736..1a065eb 100644
    --- a/lib/ipmi_main.c
    +++ b/lib/ipmi_main.c
    @@ -379,7 +379,7 @@ ipmi_main(int argc, char ** argv,
    char * seloem = NULL;
    int port = 0;
    int devnum = 0;
    - int cipher_suite_id = 3; /* See table 22-19 of the IPMIv2 spec */
    + int cipher_suite_id = 1; /* See table 22-19 of the IPMIv2 spec */
    int argflag, i, found;
    int rc = -1;
    char sol_escape_char = SOL_ESCAPE_CHARACTER_DEFAULT;

     
  • Zdenek Styblik

    Zdenek Styblik - 2012-10-11

    Hello,

    you can specify cypher suite via -C parameter. Does it work, resp. if you supply # ipmitool -C 1 ... ; does it fix problems you're having?

     
  • arekm

    arekm - 2012-10-11

    -C 3 indeed also fixes the problem. Now the question is why this is needed at all? Buggy BMC (that doesn't like cipher 1 (which is default now)) or some other reason?

     
  • Jim Mankovich

    Jim Mankovich - 2012-10-11

    There is defect associated with the change of cipher suite from 3 to 1, see http://sourceforge.net/tracker/?func=detail&aid=3571371&group_id=95200&atid=610550.
    I think we should change the default cipher suite default back to 3 since that is what is was in 1.8.11. If you want to run with 1, use the -C 1 switch. Any opinions?

     
  • Zdenek Styblik

    Zdenek Styblik - 2012-10-15

    I thought the reply is going to be easy. It isn't. Revision log says ``Changed default cipher suite to 1 instead of 3 for iol20'' by blaism. Ok, I'll ask the stupid question - what is iol20?

    I honestly don't know how to be helpful and constructive here. Prior to changing anything back, I would try to find out why it was changed in the first place.
    Second thing. Since IPMI supports ``Get Channel Cipher Suites'' command, why it isn't implemented in a way 1] if '-C' is not specified, get list of Cipher Suites 2] choose the best, the most secure, cipher suite from the list 3] connect to BMC.

     
  • Zdenek Styblik

    Zdenek Styblik - 2012-10-15

    Grh! I've just repeated Jim's comment from the other bug report.

     
  • Zdenek Styblik

    Zdenek Styblik - 2012-10-15

    iol =~ IPMI over LAN; iol20 =~ IPMI over LAN 2.0. I should have realized earlier Sorry for the spam.

    I flew through IPMI specification and I can't find reason to change it one way or another. I'm sorry I can't put more time into this specific area right now. I'm not against changing cipher suite back to 3, although I wonder what was the reason for changing it to 1.

     
  • Jim Mankovich

    Jim Mankovich - 2012-10-17

    I looked at the IPMI spec and I can find no good reason why the cipher suite id needs to be 1 for IPMI over LAN 2.0. So, I see no problem with changing the default cipher id suite back to 3.

     
  • Zdenek Styblik

    Zdenek Styblik - 2013-03-09
    • milestone: --> version-1.8.12
     
  • Zdenek Styblik

    Zdenek Styblik - 2013-03-11
    • assigned_to: nobody --> jmank
    • status: open --> closed-fixed
     
  • Zdenek Styblik

    Zdenek Styblik - 2013-03-11

    Code has been reverted, resp. back to default auth 3.

     

Get latest updates about Open Source Projects, Conferences and News.

Sign up for the SourceForge newsletter:





No, thanks