[Spread-users] Increasing the MAX_GROUP_NAME and/or the MAX_PRIVATE_NAME?

Ed Holyat eholyat at olf.com
Fri Feb 3 12:24:36 EST 2006

Hello, I have been monitoring this list for sometime and was hoping for
insight into a spread problem.

Has anyone run into issues increasing the MAX_GROUP_NAME and/or the

I am currently testing spread with a modified group and private name.
The default constraints are impractical for the usage.
#define	MAX_GROUP_NAME		255
#define     MAX_PRIVATE_NAME        100

I am experiencing random crashes in the Daemons while testing.  
I am unable to create a consistent reproducible case to diagnose the

The crash does not occur frequently. 
The Group Name + Private name average 128 bytes in my testing which is
far less than the 255 bytes that I specified.  

There are 2 issues I would like to resolve.
1.      When the crash occurs, it seems the private name is being
partially copied over by another private name (The same name), or the
private name is getting cut off  (memory corruption?).   This results in
either a SEGV or the entire Network dropping because of a bad private
name.  It always seems that it is the 2nd part of the private name
(daemonhost) being left off or getting over written.


Is the 255 byte Group name or 100 byte private name too large?  
Is it a combination of private_name + group_name that has to fit in a
byte. I believe the max private_name + group_name Has to fit into a byte
255?  Should my definitions be 
#define     MAX_PRIVATE_NAME        100

2.      The entire network drops, because of the bad private name.  Is
there a configuration parameter that will change the behavior of this
Alarm or do I have to change the code?

        [1] #AAA1090.2880.Master.xxmm.AAA1090.2880.2880.async.1#AAA1090
[462] group mmmm_smmmm10g_xxxxxxxxx.xxxxxxxxx.STATUS with 1 members:
        [1] #AAA1090.2880.yyyyyy.xxmm.AAA1090.2880.2880.async.1#AAA1090
Sess_handle_kill: Illegal private name to kill
Exit caused by Alarm(EXIT)

More information about the Spread-users mailing list