Jeremy Chadwick koitsu at FreeBSD.org
Mon Nov 17 04:57:25 PST 2008

On Mon, Nov 17, 2008 at 02:49:26PM +0200, Ott Köstner wrote:
> Manolis Kiagias wrote:
>> Ott Köstner wrote:
>>> Manolis Kiagias wrote:
>>>> I don't think it is a matter of days, we have not even reached RC  
>>>> status yet on 7.1
>>>> On  a production server you will probably wish to go with  
>>>> 7.0-RELEASE-p5. It would be trivial to upgrade to 7.1 by means of  
>>>> freebsd-update(8) when it is released.
>>>> You probably don't want to risk 7.1-PRERELEASE on a server, but for 
>>>> anyone running workstations, desktops, laptops I think it is worth  
>>>> trying at this moment.
>>> I am a person, who made a mistake, installing 7.1 on my production  
>>> server (actually RELENG_7 stable, which shows up as 7.1).
>>> My question is, how stupid is that mistake? Is it better to reinstall 
>>> 7.0 before something really bad happens, or can I just let it run?  
>>> What are the most serious bugs to expect?
>>> Greetings,
>>> O.K.
>> It all depends on the programs you run, your configuration, system  
>> load and so on. Bugs that may be present in the system, may simply not  
>> be applicable to you, if you are not using the specific part or  
>> feature that has the problem.  While it is difficult to assess without  
>> knowing specific details, I think 7.1 is generally stable at the  
>> moment. Maybe people using it in production servers (if any) can step  
>> in and share their experiences.
> I have just regular internet things running:
> apache22, mysql5, bind94, php5, postfix, dovecot, proftpd, clamav,  
> spamassasin, snmpd...
> ...plus ipfw and pf

I'm not sure why you're paranoid of 7.1-PRERELEASE (RELENG_7 at the
present time).  It seems like you installed it or upgraded to it, then
became "greatly concerned" for strange reasons.

I really don't think you have anything to worry about.  The only
"gigantic" difference, as I see it, is that the ULE scheduler is now the
default (while in 7.0, 4BSD was default).

I personally run all of the above services (minus bind94, clamav, and
snmpd; I use the base system bind, do not care for AV software, and use
bsnmpd where applicable), and I also run all of those services on
servers in my co-location.  We have a RELENG_7 box which is working fine
doing all of the above, on a dual-core processor to boot.

> Actually there is one strange thing what I described here before --  
> 'top' reporting incorrect process times. But this appeared already when  
> I upgraded from FreBSD 6.3 to 7.0. The bug appears on all 7.++ machines,  
> i386 and amd64. Everything is OK with single-threaded processes, but  
> processes with multiple threades are reported incorrectly by top.  
> Anybody else has experienced this?

Sounds familiar (not personally).  There are efforts underway in CURRENT
to improve top.  I would recommend posting your top problem to -stable
or -hackers.

> Also, there was a strange phenomenon with FreeBSD router with pf.
> the rule in question is:
> 'scrub in all'
> I do not knw, if this has anything to do with 7.1 issue. Maybe it is not  
> just a good idea to have a 'scrub' rule on router...

No, it's perfectly fine.  But your description of the problem is too
terse, and the issue should be discussed on freebsd-pf not here.

There *are* other problems with pf which have been fixed in RELENG_7
(7.1), and you can review my Wiki to see what those are.

| Jeremy Chadwick                                jdc at parodius.com |
| Parodius Networking                       http://www.parodius.com/ |
| UNIX Systems Administrator                  Mountain View, CA, USA |
| Making life hard for others since 1977.              PGP: 4BD6C0CB |

More information about the freebsd-questions mailing list