The Ultimate BSD Thread™ - The Devil Wears a Neckbeard

The Something Awful Forums > Discussion > Serious Hardware/Software Crap > The Ultimate BSD Thread™ - The Devil Wears a Neckbeard

«‹›34 »

Computer viking: May 30, 2011; Now with less breakage.

hifi posted:

Set up ssh (if you didn't enable it in the installer) and use scp

Exactly this. On the client side, WinSCP works well on Windows (though the interface is pure 1997), and I think cyberduck is fine on a Mac - or the command line tools. On Linux I use the command line tools , but you can also do fish:// or sftp:// in any KDE program (including dolphin).

# ? Jun 6, 2014 09:25

Adbot: ADBOT LOVES YOU

# ? May 30, 2024 00:43

Lord Windy: Mar 26, 2010

I had no idea ssh existed. It is perfect for what I needed to do, I almost feel like crying that something that was giving me such grief ended up having such an elegant solution.

edit:

I'm just using Terminal that comes with the Mac

# ? Jun 6, 2014 10:47

Cpt.Wacky: Apr 17, 2005

Samba 3.x is available as a package for OpenBSD and works for basic network file shares too if you want something more convenient.

# ? Jun 6, 2014 19:15

thebigcow: Jan 3, 2001; Bully!

^^ Wouldn't NFS be easier between BSD and OSX? Or are they not quite compatible and require a bunch of work?

I like FileZilla on Windows for sftp, it even uses your key agent. I don't know how good the OSX version is though.

# ? Jun 6, 2014 19:42

evol262: Nov 30, 2010; #!/usr/bin/perl

thebigcow posted:

^^ Wouldn't NFS be easier between BSD and OSX? Or are they not quite compatible and require a bunch of work?

I like FileZilla on Windows for sftp, it even uses your key agent. I don't know how good the OSX version is though.

NFS is just terrible on OSX.

# ? Jun 6, 2014 20:12

Ninja Rope: Oct 22, 2005; Wee.

What's the best way to update openssl on a freebsd 10 machine with a custom kernel? I can't use freebsd-update because the kernel is 10-stable. The machine is not powerful enough to rebuild kernel and world in a reasonable amount of time. Is my only option to rebuild world on another machine and then overwrite everything over the existing system?

# ? Jun 22, 2014 08:57

JamesOff: Dec 12, 2002; What a frightening beast!

Do you need to rebuild world or will just /usr/src/usr.bin/openssl or similar do?

# ? Jun 22, 2014 14:18

Ninja Rope: Oct 22, 2005; Wee.

JamesOff posted:

Do you need to rebuild world or will just /usr/src/usr.bin/openssl or similar do?

It seems like everything I care about is linked against /usr/lib/libssl.7 and /usr/lib/libcrypto.7. Even if I rebuild openssl or install from pkg it installs /usr/local/lib/libssl.8 and /usr/local/lib/libssl.8 and I'm hesitant to assume .7 and .8 are compatible.

# ? Jun 22, 2014 21:30

feld: Feb 11, 2008; Out of nowhere its.....

Feldman

Don't install multiple OpenSSLs on the same machine if you can avoid it. I'd delete the one from pkg.

The instructions highly recommend you rebuild the entire world, and I understand that would be painful on this machine. Can you build it on another machine? If so, transfer or NFS mount /usr/src and /usr/obj to the target machine and then do the installworld only. This will save you from having to compile on that slower machine.

# ? Jul 3, 2014 17:07

Ninja Rope: Oct 22, 2005; Wee.

feld posted:

The instructions highly recommend you rebuild the entire world

That's the conclusion I came to also. I ended up mounting actual disks via NFS and then just letting the thing rebuild world and kernel over a few days. It didn't occur to me I could do the build on a different machine and installworld on this one but I'll remember that for next time, thank you.

If only the stock FreeBSD kernel had ALTQ enabled I could use the binary updates. Why is that not enabled by default?

# ? Jul 5, 2014 13:22

feld: Feb 11, 2008; Out of nowhere its.....

Feldman

Ninja Rope posted:

That's the conclusion I came to also. I ended up mounting actual disks via NFS and then just letting the thing rebuild world and kernel over a few days. It didn't occur to me I could do the build on a different machine and installworld on this one but I'll remember that for next time, thank you.

If only the stock FreeBSD kernel had ALTQ enabled I could use the binary updates. Why is that not enabled by default?

I don't know the history, but I'm guessing it's because ALTQ affects the network stack in some edge cases. But honestly, ALTQ is junk and I hope we kill it like OpenBSD did and bring in their updated pf with the new fancy QoS code.

# ? Jul 5, 2014 13:59

EvilMoFo: Jan 1, 2006

IPSEC isnt in there by default, either. That's the only reason I compile a kernel.

Also, the new pf would be super nice, if only for the current syntax.

# ? Jul 6, 2014 22:47

Ninja Rope: Oct 22, 2005; Wee.

How's the current maintainer of FreeBSD's pf doing? Is he able to keep up or does he need help?

# ? Jul 7, 2014 00:09

Xenomorph: Jun 13, 2001

Since setting up a new FreeBSD 10.0 server, I've been using the pkg command more and more.

However, this happens on our older FreeBSD 9.2 server:

code:

# pkg upgrade

Checking integrity... done (1 conflicting)
pkg: cannot resolve conflict between remote perl5.14(want install), remote perl5(want install), please resolve it manually
pkg: SAT: conflicting request, cannot solve
pkg: cannot solve job using SAT solver

Edit: It looks like a conflict with perl5.14 and perl5.16 (which goes by "perl5" now).

"pkg install perl5.14" reinstalls my current perl, but "pkg install perl5" wants to install perl5.16 and remove perl5.14 and everything that depended on it (samba, webmin, cowsay, etc).

Double-edit:

It looks like I may have to do something like this to replace the port and then re-compile everything using it:

code:

# portmaster -o lang/perl5.16 lang/perl5.14

Triple-edit, yeah. It looks like it was issue 20120630 in /usr/ports/UPDATING

4th edit:
That worked. The "portmaster -o" command replaced the port and left everything else alone, then "pkg upgrade" reinstalled a few things because "direct dependency changed" for them.

Xenomorph fucked around with this message at 01:30 on Jul 24, 2014

# ? Jul 23, 2014 23:35

Computer viking: May 30, 2011; Now with less breakage.

Ninja Rope posted:

How's the current maintainer of FreeBSD's pf doing? Is he able to keep up or does he need help?

There's a discussion in the -current mailing list about importing a more current version; so far no one has volunteered. Bapt looked at it and decided it's ... complicated, since the current version has things like multithreading that aren't in the OpenBSD version.

# ? Jul 24, 2014 00:37

feld: Feb 11, 2008; Out of nowhere its.....

Feldman

pf: extremely popular, no maintainer, some annoying quirks. new users tend to use this.
ipf: haha, who uses this anyway? it's the reason pf exists.
ipfw: everyone old school uses this, and supposedly it works perfectly. I hate the config syntax though. Luigi Rizzo seems to be the current maintainer.

edit: not be so harsh on pf

feld fucked around with this message at 17:07 on Jul 24, 2014

# ? Jul 24, 2014 16:41

Marinmo: Jan 23, 2005; Prisoner #95H522 Augustus Hill

I'm having terrible network performance between my windows computer and my freebsd server.

code:

[root@KOSMOS:/home/xxx] iperf -m -c 192.168.1.2
------------------------------------------------------------
Client connecting to 192.168.1.2, TCP port 5001
TCP window size: 32.5 KByte (default)
------------------------------------------------------------
[  3] local 192.168.1.10 port 62342 connected with 192.168.1.2 port 5001
[ ID] Interval       Transfer     Bandwidth
[  3]  0.0- 6.7 sec   469 MBytes   589 Mbits/sec
[  3] MSS size 1448 bytes (MTU 1500 bytes, ethernet)
[root@KOSMOS:/home/xxx] iperf -m -s
------------------------------------------------------------
Server listening on TCP port 5001
TCP window size: 64.0 KByte (default)
------------------------------------------------------------
[  4] local 192.168.1.10 port 5001 connected with 192.168.1.2 port 58553
[ ID] Interval       Transfer     Bandwidth
[  4]  0.0- 9.7 sec  51.2 MBytes  44.5 Mbits/sec
[  4] MSS size 1460 bytes (MTU 1500 bytes, ethernet)

The card in question is an onboard marvell yukon chipset.

code:

msk0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> metric 0 mtu 1500
        options=c011a<TXCSUM,VLAN_MTU,VLAN_HWTAGGING,TSO4,VLAN_HWTSO,LINKSTATE>
        ether 00:50:8d:b2:98:d1
        inet 192.168.1.10 netmask 0xffffff00 broadcast 192.168.1.255
        inet6 fe80::250:8dff:feb2:98d1%msk0 prefixlen 64 scopeid 0x1
        nd6 options=29<PERFORMNUD,IFDISABLED,AUTO_LINKLOCAL>
        media: Ethernet autoselect (1000baseT <full-duplex,flowcontrol,rxpause,txpause>)
        status: active

code:

mskc0@pci0:4:0:0:       class=0x020000 card=0x1084147b chip=0x436411ab rev=0x13 hdr=0x00
    vendor     = 'Marvell Technology Group Ltd.'
    device     = '88E8056 PCI-E Gigabit Ethernet Controller'
    class      = network
    subclass   = ethernet

Anyone else run into problems with this particular combination? As can be seen; the problem arise when connecting from windows computer to the freebsd box. I notice the TCP window size is different, can that be a factor? When I try to lower the window size to 32kb, when quitting the program on the windows machine, the thread seem to hang and I never get any results from iperf. This is, obviously, over a GB network, so even the ~5-600mbps I get when the connection is good seems a tad bit low to be honest.

# ? Aug 20, 2014 01:00

unknown: Nov 16, 2002; Ain't got no stinking title yet!

Your network card (driver?) is crap. When you do the ifconfig, you can see on the options line that it only supports txcsum (transmit check sum offloading) and not receive offload.

So your cpu is going nuts doing all the work which entails a huge performance hit as you can see from your tests.

# ? Aug 20, 2014 02:18

Marinmo: Jan 23, 2005; Prisoner #95H522 Augustus Hill

unknown posted:

Your network card (driver?) is crap. When you do the ifconfig, you can see on the options line that it only supports txcsum (transmit check sum offloading) and not receive offload.

So your cpu is going nuts doing all the work which entails a huge performance hit as you can see from your tests.

Ah, gotcha. So just switching to another network card (I have plenty in this thing) would hopefully give me a better result then. I'll try and get back to you, thanks!

# ? Aug 20, 2014 12:20

Marinmo: Jan 23, 2005; Prisoner #95H522 Augustus Hill

Followup:

Switching to another network card ... It became better, but considering 1gbps, there's no other judgement to make than "abysmal".

code:

------------------------------------------------------------
Server listening on TCP port 5001
TCP window size: 64.0 KByte (default)
------------------------------------------------------------
[  4] local 192.168.1.10 port 5001 connected with 192.168.1.2 port 56928
[ ID] Interval       Transfer     Bandwidth
[  4]  0.0-10.1 sec  99.6 MBytes  83.0 Mbits/sec
[  4] MSS size 1460 bytes (MTU 1500 bytes, ethernet)
[  5] local 192.168.1.10 port 5001 connected with 192.168.1.2 port 56929
[  5]  0.0-10.1 sec   135 MBytes   112 Mbits/sec
[  5] MSS size 1460 bytes (MTU 1500 bytes, ethernet)

code:

------------------------------------------------------------
Client connecting to 192.168.1.2, TCP port 5001
TCP window size: 32.5 KByte (default)
------------------------------------------------------------
[  3] local 192.168.1.10 port 61307 connected with 192.168.1.2 port 5001
[ ID] Interval       Transfer     Bandwidth
[  3]  0.0-10.0 sec   293 MBytes   246 Mbits/sec
[  3] MSS size 1448 bytes (MTU 1500 bytes, ethernet)
------------------------------------------------------------
Client connecting to 192.168.1.2, TCP port 5001
TCP window size: 32.5 KByte (default)
------------------------------------------------------------
[  3] local 192.168.1.10 port 29021 connected with 192.168.1.2 port 5001
[ ID] Interval       Transfer     Bandwidth
[  3]  0.0-10.2 sec   252 MBytes   208 Mbits/sec
[  3] MSS size 1448 bytes (MTU 1500 bytes, ethernet)

code:

re1: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> metric 0 mtu 1500
        options=8209b<RXCSUM,TXCSUM,VLAN_MTU,VLAN_HWTAGGING,VLAN_HWCSUM,WOL_MAGIC,LINKSTATE>
        ether f8:1a:67:03:99:31
        inet 192.168.1.10 netmask 0xffffff00 broadcast 192.168.1.255
        inet6 fe80::fa1a:67ff:fe03:9931%re1 prefixlen 64 scopeid 0x4
        nd6 options=29<PERFORMNUD,IFDISABLED,AUTO_LINKLOCAL>
        media: Ethernet autoselect (1000baseT <full-duplex>)
        status: active

code:

re1@pci0:5:5:0: class=0x020000 card=0x816910ec chip=0x816910ec rev=0x10 hdr=0x00
    vendor     = 'Realtek Semiconductor Co., Ltd.'
    device     = 'RTL8169 PCI Gigabit Ethernet Controller'
    class      = network
    subclass   = ethernet

Most everything is default on both machines as far as network configuration goes. Had win 2012 server on the server before, where it comfortably achieved ~90MB/s throughput on standard file transferring.

# ? Aug 26, 2014 12:10

Volguus: Mar 3, 2009

Speaking of weird iperf speeds:

code:

------------------------------------------------------------
Client connecting to 192.168.1.1, TCP port 5001
TCP window size: 32.3 KByte (default)
------------------------------------------------------------
[  3] local 192.168.1.1 port 4621 connected with 192.168.1.1 port 5001
[ ID] Interval       Transfer     Bandwidth
[  3]  0.0-10.0 sec  10.3 GBytes  8.85 Gbits/sec
[  3] MSS size 33092 bytes (MTU 33132 bytes, unknown interface)

Card:

code:

em1: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500                                                                                                                                     
        lladdr 40:16:7e:28:2a:bc                                                                                                                                                                     
        priority: 0                                                                                                                                                                                  
        media: Ethernet autoselect (1000baseT full-duplex,rxpause,txpause)                                                                                                                           
        status: active

Something looks fishy to me with that 8.85 GBits/sec, but hey ... they work

This is OpenBSD 5.5

# ? Aug 26, 2014 13:20

Bluecobra: Sep 11, 2001; The Future's So Bright I Gotta Wear Shades

Marinmo posted:

Followup:

Switching to another network card ... It became better, but considering 1gbps, there's no other judgement to make than "abysmal".

Have you tried running UDP iperf tests instead? That would rule out any TCP tunables.

# ? Aug 27, 2014 05:10

Marinmo: Jan 23, 2005; Prisoner #95H522 Augustus Hill

Bluecobra posted:

Have you tried running UDP iperf tests instead? That would rule out any TCP tunables.

With UDP transfers seem to cap at 1mbit/s for some reason. :ohdear:

Perhaps I'd be best off buying only Intel NICs if I'm going to do this in the future ... Since the cards in the machine now work just dandy w/ both windows/linux (dislike win2012 server, hate the ideology behind systemd if you wonder why I don't use 'em), I'm willing to, perhaps prematurely, call this a driver issue w/ Realtek chips as well?

# ? Aug 29, 2014 22:29

SamDabbers: May 26, 2003

Marinmo posted:

With UDP transfers seem to cap at 1mbit/s for some reason.

Perhaps I'd be best off buying only Intel NICs if I'm going to do this in the future ... Since the cards in the machine now work just dandy w/ both windows/linux (dislike win2012 server, hate the ideology behind systemd if you wonder why I don't use 'em), I'm willing to, perhaps prematurely, call this a driver issue w/ Realtek chips as well?

You need to tell iperf how fast to send UDP packets. Unlike TCP, it doesn't have a mechanism to determine if packets are being dropped. The option is -b:

code:

iperf -b 8M  # 8Mbps
iperf -b 600K  # 600Kbps

# ? Aug 29, 2014 22:32

Ninja Rope: Oct 22, 2005; Wee.

If I have a host that is running a custom kernel, but is almost completely stock 10-STABLE, is there a way to "update" it to a version from freebsd-update? I know I'll lose whatever I did differently building my own custom kernel, but they're so similar I don't think it will matter. I'd rather get back on binary releases.

# ? Aug 30, 2014 13:29

EvilMoFo: Jan 1, 2006

10-stable isn't a tracked branch for freebsd-update, unless something changed. I don't know if it would even take kindly do you specified a release patch level to update to.

I remember reading that there is some magical way to avoid freebsd-update nuking your custom kernel, so long as the base kernel is named generic or some such. I haven't really looked too closely into it though.

# ? Aug 30, 2014 20:47

Xenomorph: Jun 13, 2001

What am I missing here?

code:

# portmaster net/samba41

===>>> Port directory: /usr/ports/net/samba41

===>>> Gathering distinfo list for installed ports

===>>> Launching 'make checksum' for net/samba41 in background
===>>> Gathering dependency list for net/samba41 from ports
===>>> Initial dependency check complete for net/samba41


===>>> Starting build for net/samba41 <<<===

===>>> All dependencies are up to date

===>  Cleaning for samba41-4.1.11
===>  License GPLv3 accepted by the user
===>  Found saved configuration for samba41-4.1.11
===>   samba41-4.1.11 depends on file: /usr/local/sbin/pkg - found
===> Fetching all distfiles required by samba41-4.1.11 for building
===>  Extracting for samba41-4.1.11
=> SHA256 Checksum OK for samba-4.1.11.tar.gz.
===>  Patching for samba41-4.1.11
===>  Applying FreeBSD patches for samba41-4.1.11
===>   samba41-4.1.11 depends on package: p5-Parse-Pidl>=4.0.0 - found
===>   samba41-4.1.11 depends on package: talloc>=2.1.0 - found
===>   samba41-4.1.11 depends on package: tevent>=0.9.18 - found
===>   samba41-4.1.11 depends on package: tdb>=1.2.11 - found
===>   samba41-4.1.11 depends on package: ntdb>=1.0 - found
===>   samba41-4.1.11 depends on package: ldb>=1.1.17 - found
===>   samba41-4.1.11 depends on executable: gdb771 - not found
===>    Verifying install for gdb771 in /usr/ports/devel/gdb
===>  Installing for gdb-7.8_2
===>   gdb-7.8_2 depends on shared library: libreadline.so.6 - found (/usr/local/lib/libreadline.so.6)
===>  Checking if gdb already installed
===>   gdb-7.8_2 is already installed
      You may wish to ``make deinstall'' and install this port again
      by ``make reinstall'' to upgrade it properly.
      If you really wish to overwrite the old port of gdb
      without deleting it first, set the variable "FORCE_PKG_REGISTER"
      in your environment or the "make install" command line.
*** Error code 1

Stop.
make[3]: stopped in /usr/ports/devel/gdb
*** Error code 1

Stop.
make[2]: stopped in /usr/ports/devel/gdb
*** Error code 1

Stop.
make[1]: stopped in /usr/ports/net/samba41
*** Error code 1

Stop.
make: stopped in /usr/ports/net/samba41

===>>> make build failed for net/samba41
===>>> Aborting update


===>>> You can restart from the point of failure with this command line:
       portmaster <flags> net/samba41 
#

I'm guessing it's looking for gdb 7.7 ("Verifying install for gdb771"), when gdb 7.8 is current. I've deleted & rebuilt both. /usr/ports/devel has gdb 7.8 and gdb 6.6.

What do I need to fix to get samba41 to install? Do I need to get a hold of gdb771 somehow?

Edit:
It looks like if I re-compile samba41 without the "developer" option, it doesn't need gdb7.7.

Why would Samba require "gdb771" exact instead of ">=gdb771" like all other dependencies?

Xenomorph fucked around with this message at 00:41 on Sep 19, 2014

# ? Sep 19, 2014 00:35

EvilMoFo: Jan 1, 2006

Have you emailed the port maintainer?

# ? Sep 19, 2014 15:04

Xenomorph: Jun 13, 2001

EvilMoFo posted:

Have you emailed the port maintainer?

The FreeBSD net/samba41 maintainer? Nope. I didn't think I could do that. How do I do that?

# ? Sep 19, 2014 18:57

thebigcow: Jan 3, 2001; Bully!

http://www.freshports.org/net/samba41/

# ? Sep 19, 2014 19:07

Xenomorph: Jun 13, 2001

Another fun thing with Samba 4/4.1. Apparently winbind cannot read uidNumber correctly from AD any more...?

I have this in my smb.conf (Samba 3 on FreeBSD 9.2) and smb4.conf (Samba 4.1 on FreeBSD 10.0):

code:

# AD objects without a UID/GID will be mapped to these
idmap config * : range = 50000-59999
idmap config * : backend = tdb

# domain-specific, get info from AD
idmap config DOMAIN : schema_mode = rfc2307
idmap config DOMAIN : backend = ad
idmap config DOMAIN : range = 10000-19999

Samba 3:

code:

$ wbinfo -i derp
derp:*:1000:1000:MyName:/home/derp:/bin/false

(correct)

Samba 4:

code:

$ wbinfo -i derp
derp:*:50000:50000:MyName:/home/derp:/bin/false

(nope!)

On both systems, using LDAP/nsswitch:

code:

$ getent passwd | grep derp
derp:*:1000:1000:MyName:/home/derp:/bin/bash

Some searches said try compiling with "--with-shared-modules=idmap_ad", can that be done with portmaster? I've read that I should just give up on winbindd and use sssd...?

I've read about the new "idmap_ldb:use rfc2307 = yes" option. But that doesn't seem to help.

It seems like a pretty big thing to just allow to break.

Windows AD, FreeBSD/Samba Clients

I think this person may have a similar issue:
https://bugzilla.samba.org/show_bug.cgi?id=9795

I may just wipe Samba 4.1 and put Samba 3.6 on FreeBSD 10.0. I don't think I'm getting any advantages from using 4/4.1, and I don't want to bother if random things seem broke with it.

# ? Sep 24, 2014 00:06

Xenomorph: Jun 13, 2001

Man, this software has been a bucket of poo poo.

FreeBSD 10.0, Samba 4.1/Winbind just won't work. Active Direction has supported standard UNIX uidNumber/gidNumber since at least Windows Server 2003 R2. The weird "idmap" function in Samba 3.6 at least work (why would you need to "map" anything to a range? Just read the number as-is in AD).

So I remove Samba 4.1 and install Samba 3.6... But that doesn't work on FreeBSD 10.0!

winbind crashes with a core dump and the "sha1 checksum failed" message.

"Critical" bug opened in July of 2013, and zero progress has been made:
https://bugzilla.samba.org/show_bug.cgi?id=10039
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=186694

So I'm guessing I have to just stay on FreeBSD 9 and Samba 3 forever?

The FreeBSD roadmap lists support for 9.3 until the end of 2016.

The Samba people seemed to drop 3.6 and 4.0 and are just doing development on 4.1.

Is what I'm doing really that so foreign?

* Windows Active Directory (users, authentication, etc).
* FreeBSD NFS/Samba server (authenticated to AD).

NFS/LDAP works fine. All users have the correct uidNumber and gidNumber.
Samba/Winbind has been a pain.
Samba own documentation doesn't work (https://wiki.samba.org/index.php/Samba_&_Active_Directory). Are their docs for Samba 3?

The Samba 4 Winbind page has been deleted: https://wiki.samba.org/index.php/Samba4/Winbind

Xenomorph fucked around with this message at 03:51 on Sep 26, 2014

# ? Sep 26, 2014 03:41

Cpt.Wacky: Apr 17, 2005

I never had much luck with samba beyond basic file sharing on a home network with version 2 and 3 on OpenBSD. I specifically fought quite a lot with version 3 and then 4 on getting the LDAP backend to work and gave up. They don't seem to care at all about supporting platforms other than chosen linux distros.

# ? Sep 26, 2014 19:27

Ninja Rope: Oct 22, 2005; Wee.

I'm using Samba 4 on FreeBSD 10 just fine, but I built it from source and I'm not using UID mapping or anything. Users have a unix account they log in as.

# ? Sep 26, 2014 22:07

Xenomorph: Jun 13, 2001

Our setup cannot be that unique.

We have Active Directory for centralized user management. Isn't that like, an industry norm?

I just want a file server (that needs to provide Samba and NFS to clients) that authenticates to our AD server. It's been working fine for 2+ years on our Samba3/FreeBSD9 server.

Samba4 isn't feature-complete, yet, but they still seemed to have dropped Samba3. Only Samba 4.1 is listed as a current project.

Samba3/Winbind works on FreeBSD 9, but has weird encryption and crashing errors on FreeBSD 10. Is that a Samba issue or FreeBSD issue? Would some libraries change in 10.0 that Samba 3 relied on? Or does FreeBSD 10.0 use different Samba3 sources than FreeBSD 9?

Samba4/Winbind cannot seem to query AD for UID/GID information, and that seems like quite an oversight. Security on Linux and UNIX is all about the UID/GID, and I just cannot get Samba4 to read that from AD.

# ? Sep 27, 2014 02:02

Computer viking: May 30, 2011; Now with less breakage.

I don't have any control over the ancient AD domain i authenticate against, so I'm using the uid mapping. As long as the username/uid mapping is stable, it should be fine - the only issue I had with differing uids between the machines was the nfs3-mounted file server.

Moving to nfs4 solved that, though that's not absolutely trivial either. Or, to be exact: it seems to work fine without being to onerous to set up now, with a recent freebsd serving ubuntu 14.04 clients. With earlier versions it was a deep well of things that should, but didn't, work. (I haven't set up sec=krb5 yet, though. That looks like it'll be "fun".)

# ? Sep 27, 2014 09:18

Nam Taf: Jun 25, 2005; I am Fat Man, hear me roar!

I've got a raid-z pool of 4 drives plus a single UFS drive. What load would the UFS drive see by mounting the zpool and sharing it over LAN? Am I right in understanding that there would be minimal caching on the drive, instead using my RAM?

I'm sick of having my OS drive fail as I'm onto my 2nd failure in a few months and so I'm thinking I'll just grab a USB thumb drive and boot from that. The boot time isn't too big a problem and it usually just sits there serving files off the Zpool.

What capacity would I want for a basic FreeBSD 10 install with stuff like samba, etc. configured? 32GB?

# ? Sep 28, 2014 06:14

Xenomorph: Jun 13, 2001

Nam Taf posted:

I've got a raid-z pool of 4 drives plus a single UFS drive. What load would the UFS drive see by mounting the zpool and sharing it over LAN? Am I right in understanding that there would be minimal caching on the drive, instead using my RAM?

The "rule of thumb" I've seen is 1GB of RAM for every 1TB of storage.

Although, 1GB per 1TB may not be enough. My pair of FreeBSD servers have 64GB of RAM and 36TB of storage, with ~24TB of usable storage (two raidz2s striped, and a pair of UFS drives in hardware RAID 1 for the OS drive), and I've maxed out the RAM before on them several times. It's sitting idle right now, with about 28GB of memory in use, but during the middle of the day it has no free memory and it starts hitting the swap partition (16GB) pretty hard.

I'm going with the default ZFS compression level, and no data deduplication (increasing compression level or going with deduplication ate tons of CPU and slowed down network transfers).

Have you considered going with an SSD for your OS drive? I'd recommend the Intel 530 SSD, 120GB or 240.

I have a full install of FreeBSD 9.2 (all options selected at install), all the port sources downloaded, and then Samba 3.6 installed. It's using about 11GB of my OS drive.

# ? Sep 30, 2014 02:53

Captain Pike: Jul 29, 2003

I wanted to see what it is like to run the GnuStep desktop environment, so I installed MidnightBSD in Virtual Box. http://www.midnightbsd.org/

You have to compile X before you can use it.

X doesn't compile on it.

This all made me very sad.

edit: Oh.. It looks like what I really want is WindowMaker!

Captain Pike fucked around with this message at 07:46 on Oct 1, 2014

# ? Oct 1, 2014 07:38

Adbot: ADBOT LOVES YOU

# ? May 30, 2024 00:43

alyandon: Dec 9, 2001; Poster of the Month for July!; Fun Shoe

Xenomorph posted:

Our setup cannot be that unique.

We have Active Directory for centralized user management. Isn't that like, an industry norm?

Nope, not at all unique. We have a random assortment of FreeBSD machines (v8 through v10) for some very specific tasks among a seemingly endless sea of RHEL/Centos/Windows boxes that are all tied into Windows AD for authentication and authorization.

I'll ask someone in our unixsys group about how they are pulling off the AD integration on our FreeBSD boxes and update the thread if I get a coherent answer.

Edit: Turns out we aren't using samba on those servers. Sorry.

alyandon fucked around with this message at 20:45 on Nov 13, 2014

# ? Nov 12, 2014 02:34

The Something Awful Forums > Discussion > Serious Hardware/Software Crap > The Ultimate BSD Thread™ - The Devil Wears a Neckbeard

«‹›34 »