Re: A novel idea....

2001-04-05 Thread sean-freebsd-hackers

Read #1, but skip the rest, it's opinion.



BSD license, right?  Without disagreeing with any of the
previous points, let me step into evangelism mode here and borrow/add
my own comments.  Please take replies to the evangelism list, however
(maybe post a variant of this on FreeBSD.org someplace).

 In short, these are the reasons I prefer FreeBSD:

1) The file system is so much better than any Linux file system,
ReiserFS included.

Question: Is UFS a balanced btree?  I don't think so, but I
could be wrong.  In either case, UFS is by far and away superior to
ext[\d].  Check out soft-updates if you really want to be impressed.


2) The system isn't made by idiots.

This is opinion.  There are some bright people in the Linux
crowd, but there are also some "hackers" that hack kludges instead of
elegantly solving the problem.  I've found, in my experience, that the
FreeBSD development team seems to search out the elegant/correct
solution as opposed to the quick solution.  Idiots may be going a bit
far and the start of a flame war (please avoid this, or let me know if
you want a list for flames and I'll set one up, but please not here!).

3) The system's development is controlled, and the system is
consistent because of that.

This is one of the great benefits of a central development
model/system.

4) FreeBSD never trashed my data.

I 2nd this.  I've lost many a GB of data to ext2 or other
Linux FS's.  UFS has been a rock for many years and many many many
many servers, and a long time ago in a galaxy far far away, I think I
heard a rumor of lost data... but that was the HD spindle coming loose
and shredding the platter (ie, not something the file system could do
much about).

7) I hate RMS with a passion (remember, he's the Communist hypocrite
who claims his software is Free).

I shouldn't comment, so I won't.  ::grin::

8) For a firewall, ipfw blows the doors off of Linux's
iptables/ipchains/ipmasq/whatever.

If you want a stateful firewall, look at ipf (also very very
very very very nice!!!)  I'm waiting for ipf to get bridging support
in the kernel, then you'd have a firewall that would surpass any of
your wildest dreams (no MAC addresses on the Ethernet cards, while
retaining ipf stateful filtering).

9) I prefer the file system hierarchy.

Designed extremely well...  try an upgrade with cvsup, make
world, and mergemaster: you'll never want to administer anything else
ever again (except possibly AIX, but that maybe hardware envy on my
part).

10) Bug fixes and development happen much quicker.

See tonight and the ntpd bug.  The patch was submitted before,
or less than one hour after it hit bugtraq.

11) None of those shitty SVR4 bootscripts and symlinks; no abundance
of pointless runlevels.

rc scripts are centralized and convenient, but this is largely
SA opinion.  RC scripts are extremely easy to update, tweak, IMHO.

13) The FreeBSD base system behaves better than any Linux base system
(e.g., the stuff in /usr/bin and /bin).

Agreed, and it runs very well on old klunker systems are great
w/ FreeBSD (P100's make great bridge firewalls).  A new Linux install,
on the other hand, is typically a monolithic beast that's rather large
(disk and ram).

15) Development is more conservative (e.g., I don't see a bunch of
EXPERIMENTAL warnings in /sys/i386/conf/LINT, like I do in Linux
kernels).

If you want bleeding, on the other hand, check out -CURRENT,
which gets messages every now and then that run along the lines of
"I'm going to break such and such for a few hours while I apply some
patches, hold on."  At the same time...  I have yet to have a problem
with a morning compile of -CURRENT (I know, I'm lucky).

16) FreeBSD is lighter than Linux.

I'll leave this in.  See #13.

18) Ports.  'ya can't forget them.

19) Kernel configs are cake

20) Multi-staged booting.  You don't need to change your MBR when you
install a new kernel (or want to roll back to a different kernel).  I
think I've only been stuck high and dry w/o a bootable system twice in
four years and well over 400+ servers.

Just my ramblings.  I don't evangelize much, but it strikes me
as odd that some of this info isn't on the homepage of FreeBSD. FWIW -sc

-- 
Sean Chittenden

 PGP signature


Easy way to compute memory stats? (procfs?)

2001-03-19 Thread sean-freebsd-hackers

Is there an easy way (from script ideally) to get the following
stats:

free physical mem (avail ram)
free swap
total avail mem

any two of the three would be great.  If such a beast doesn't
exist, what are the easiest calls to use to get at them so I could
write some programs that would output this info.  -sc

-- 
Sean Chittenden[EMAIL PROTECTED]

 PGP signature


Re: Easy way to compute memory stats? (procfs?)

2001-03-19 Thread sean-freebsd-hackers

 : Is there an easy way (from script ideally) to get the following
 :stats:
 :
 :free physical mem (avail ram)
 
 This is going to be quite small on any busy machine, or machine that
 has a reasonable uptime.  The VM system will cache things unless
 there's a demand for memory.  vm.stats.vm.v_free_count has the value
 in it, but quite often will be quite a bit lower than the amount of
 memory that would be available if the system were under memory
 pressure.  If you look at top's output, there is a value labled
 cache.  The pages in this queue are clean, and can be discarded
 without needing to write them to backing store.  The number the
 system tells you isn't very useful without knowing what the system
 is doing.

Drat.  I was hoping to arrive at a number similar to top's
Free column.  Right now the vm stat from sysctl is about 25% of what
top's reporting (256MB physical).  At the same time...  the free pages
may be a good value for me to toss around given that I'm also
including the load and some statistics of the host's past.

 :free swap
 
 pstat(8) will tell you this

Duh, thanks.

 :total avail mem
 
 Do you mean physical memory, in which case the value of hw.physmem will
 tellyou.  

Phys mem + swap, which I think I can calc pretty easily now
that I have swap. -sc

-- 
Sean Chittenden[EMAIL PROTECTED]

 PGP signature