Operating System - HP-UX
1830973 Members
2408 Online
110018 Solutions
New Discussion

Random Number generator for use with sendmail

 
David L Brewster
Frequent Advisor

Random Number generator for use with sendmail

I am working on using the random number generator with the new version of sendmail (version 8.13.3)

This version has TLS support and that's why I need it...

I am getting an error saying that the random number generator is not seeded.
I don't know what that means.

Here is the full log...
Apr 11 07:45:17 hpov sm-mta[9430]: k3BBjHxC009430: SYSERR(root): dbm map "access": unsafe map file /etc/mail/access
Apr 11 07:45:17 hpov sm-mta[9430]: ruleset=check_relay, arg1=pandora, arg2=199.231.8.17, relay=pandora [199.231.8.17], reject=451 4.3.0 Temporary system failure. Please try again later.
Apr 11 07:55:17 hpov /sbin/init.d/sendmail[9518]: #### Starting sendmail ####
Apr 11 07:55:18 hpov sendmail[9536]: alias database /etc/mail/aliases rebuilt by root
Apr 11 07:55:18 hpov /sbin/init.d/sendmail[9518]: /etc/mail/aliases: 7 aliases, longest 9 bytes, 88 bytes total
Apr 11 07:55:18 hpov sendmail[9536]: /etc/mail/aliases: 7 aliases, longest 9 bytes, 88 bytes total
Apr 11 07:55:18 hpov sm-mta[9539]: starting daemon (@(#)Sendmail version 8.13.3 - Revision 2.001 - 2005/02/12): SMTP+queueing@00:30:00
Apr 11 07:55:18 hpov sm-mta[9539]: STARTTLS=server: file /etc/mail/certs/key.pem unsafe: Group readable file
Apr 11 07:55:18 hpov sm-msp-queue[9541]: STARTTLS: RAND_egd(/dev/random) failed: random number generator not seeded
Apr 11 07:55:18 hpov sm-msp-queue[9542]: starting daemon (@(#)Sendmail version 8.13.3 - Revision 2.001 - 2005/02/12): queueing@00:30:00
8 REPLIES 8
A. Clay Stephenson
Acclaimed Contributor

Re: Random Number generator for use with sendmail

Have you installed the Strong Random Number Generator? It's only available at HP-UX 11.11 and up.

http://h20293.www2.hp.com/portal/swdepot/displayProductInfo.do?productNumber=KRNG11I

It's also possible that the product is installed but the startup script, /sbin/init.d/krng has not been run.
If it ain't broke, I can fix that.
Jeff Schussele
Honored Contributor

Re: Random Number generator for use with sendmail

Hi David,

What OS are you running?
Is it 11.11 (11i)?
If so do you have the KRNG11i product installed?
A "pure" RNG was not included w/11i - you had to install KRNG11i to get it.

HTH,
Jeff
PERSEVERANCE -- Remember, whatever does not kill you only makes you stronger!
Jeff Schussele
Honored Contributor

Re: Random Number generator for use with sendmail

I must learn to type faster...Clay always beats me - but 2 seconds....durn it....is embarassing!
Cheers Clay %~))

NPP,
Jeff
PERSEVERANCE -- Remember, whatever does not kill you only makes you stronger!
TOM BROCCOLETTI
Occasional Advisor

Re: Random Number generator for use with sendmail

Restarted the rng script and got the same error.

the operating system is a 11.11 64 bit server.
Jeff Schussele
Honored Contributor

Re: Random Number generator for use with sendmail

Hi (again) Tom,

Is this the "new" 8.13.3 that HP just released for the sendmail CERT?
If so, I would report this ASAP to HP because these "unofficial" CERT patches get nowhere near the QA that "normal" patches do.

Rgds,
Jeff
PERSEVERANCE -- Remember, whatever does not kill you only makes you stronger!
A. Clay Stephenson
Acclaimed Contributor

Re: Random Number generator for use with sendmail

Well, let's find out if the random device node is functional.

dd if=/dev/random bs=256 count=1 2>/dev/null | xd -Ad -txS


You should see 128 random hex values displayed.
If it ain't broke, I can fix that.
TOM BROCCOLETTI
Occasional Advisor

Re: Random Number generator for use with sendmail

This is the new 8.13.3 version that has TLS support. I did report this to HP and they have notified the lab, however, I just wondered if anyone else has hit this problem do to some configuration type of problem.

The dd using /dev/random did produce 128 hex numbers.....
A. Clay Stephenson
Acclaimed Contributor

Re: Random Number generator for use with sendmail

I just looked at the source code and it looks as though the developer's are using the wrong #define's.
If it ain't broke, I can fix that.