Re: Mount order for ZFS, jails, and nullfs

[Torbjorn Granlund]

krad  writes:

  use zfs mount/umount
  
How?

The "zfs mount" command does not work anyting like the mount(8) command
but for ZFS volumes.  The "zfs mount" mounts a volume using its
mountpoint property.  But if that's set to "legacy", then zfs mount will
hardly work, since it cannot take a mount point parameter.

The real problem is how to make all types of mounts to be made in a
top-to-bottom order.  Even if 'zfs mount' would work like I think you
assume it works, that would not help, as far as I can tell.

-- 
Torbjörn
___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "freebsd-questions-unsubscr...@freebsd.org"

Mount order for ZFS, jails, and nullfs

[Torbjorn Granlund]

In jails, I'd like a local ZFS /, a read-only nullfs-mounted /usr, and a
local /usr/local.  (I'd also have read-only nullfs-mounted /bin, /lib,
/libexec, but let's forget about that for now.)

This way, I can upgrade the master /usr once, in one place, and have all
jails inherit it.  And my dear jail inmates can install anything in
/usr/local (such as their favourite packages/ports).

I.e., things should look like this:

/myjail/zfs
/myjail/usr nullfs ro
/myjail/usr/local   zfs

There is no problem to make this happen by issuing a handful of commands
manually after boot, but I cannot seem to get it to work automatically,
with existing boot mechanism.  The problem is that the mount of /usr
will be attempted before ZFS has mounted /myjail, the jail's root.

ZFS maintains its own mount table.  It is possible to disable the
automated mounting in ZFS by specifying the pseudo mountpoint "legacy",
and then--according to the FreeBSD manual--mount it with mount(8).
Unfortunately, FreeBSD's mount seems unable to perform mounts of ZFS
volumes.

How can I achieve the result I want after a reboot?


(The reply "use ezjail!" is not the right one.  :-) It does not do what
I want to do.  I dislike its symlink forest, and that it creates just
one ZFS filesystem for the entire jail.)

-- 
Torbjörn
___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "freebsd-questions-unsubscr...@freebsd.org"

Re: Need supported SAS controller

[Torbjorn Granlund]

Matthias Gamsjager  writes:

  >
  > The Supermicro controllers listed by FreeBSD as supported seem to be
  > rebranded Adaptec controllers, and they are therefore also disqualified.
  >
  
  The supermicro usas-l8i
  http://www.supermicro.com/products/accessories/addon/AOC-USAS-L8i.cfm
  uses a LSI chip (LSISAS 1068E SAS controller ) and works great under
  FB8.1
  
Thanks.

Unfortunately, these cards fit just in supermicro motherboards, since
they have they are reversed/mirrored compared to normal PCIe cards.

(I have a Tyan S8005.)

-- 
Torbjörn
___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "freebsd-questions-unsubscr...@freebsd.org"

Need supported SAS controller

[Torbjorn Granlund]

Is the any PCIe SAS controller that is supported by FreeBSD 8.1 that I
can purchase today?  The controller should preferably be low-profile.
RAID is not needed since I have just one disk, but if only RAID
controllers are available, I can live with that.

FreeBSD supports several controllers, but most of them seems old and
impossible to purchase today.

I have tried a new Adaptec 2405 but I have serious issues with data inte-
grity (the beginning of the virtualised disk does not read back the same
data as just written).  It is unknown if this is a problem with the cont-
roller or with the FreeBSD drivers.  Due to technical and administrative
experiences with Adaptec during this process, no other Adaptec controller
will be considered, see .

The Supermicro controllers listed by FreeBSD as supported seem to be
rebranded Adaptec controllers, and they are therefore also disqualified.

The LSI controllers FreeBSD list as supported at
 (with the mpt driver)
are no longer in production, as far as I can tell.  My hope is that some
currently produced LSI controllers actually work with FreeBSD 8.1.

Please help if you know something about that.

-- 
Torbjörn
___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "freebsd-questions-unsubscr...@freebsd.org"

Failure to use Adaptec 2405 with FreeBSD 8.1

[Torbjorn Granlund]

I am trying to get an Adaptec 2405 to work in *any* mode with FreeBSD
8.1.  I have installed FreeBSD hundreds of times, usually quiet
successfully.

I have just one disk, a Seagate Savvio 15k 146 GB SAS.  (The reason I
use a RAID controller is that it appears that FreeBSD does not support
any plain SAS controllers.)

I have tried configuring the controller for JBOD, and many types of
RAID modes.

In all tried RAID modes, FreeBSD's installation program allows me to put
FreeBSD on the disk.  When I reboot, the system hang or crashes during
initial kernel load.  (Whether it hangs or crashes seem to depend on the
size of created partitions, and other such magic.)

I then boot things in "fixit" mode.  I can then newfs any partition
without problems.  I can also mount any newly newfs'ed partition except
the first one on the disk.  You read it right, newfs succeeds but then
the partition cannot be mounted.  That strange behaviour seems
reproducible with any sort of array configuration.

Finally, I gave up array configs.  After all, I have just one disk, so
any array is somewhat degenerate.  And even if I could get it to appear
to work, I wouldn't want to put my files on a disk subsystem that seems
very buggy.

In JBOD mode, FreeBSD's installation program does not allow installation
to the disk.  (As a sanity check of the hardware and firmware, I tried
installing Debian GNU/Linux, and it works.  Using Debian is not an
option, I need FreeBSD.)

I again start fixit mode to check dmesg.  And the disk is really
there!  It shows up as a "Fixed Uninstalled SCSI 5" disk.  The
/dev/aac* devices are in place.

I am stuck at this point.  I have tried google for help.  I have read
some FreeBSD kernel sources.  I have no idea why FreeBSD rejects
installation to the disk, and what FreeBSD means with that the disk is
"Uninstalled".

System summary:

Controller: Adaptec 2405 RAID SAS PCIe
Disk:   Seagate Savvio 15k 146 GB SAS
Motherboard:ASUS M2V-MX AM2 MicroATX PCIe
FreeBSD ver:8.1

Questions:

* Can Adaptec 2405 actually be used as a boot device with FreeBSD?

* Is the firmware and/or FreeBSD drives for 2405 in such a shape that it
  can be trusted for a critical system?  Are my problems due to bugs, or
  am I trying something that is not (yet) supported.

* If Adaptec 2405 is not good with FreeBSD, what controller should I
  purchase instead, to allow me to use the new disk?  (I do not care
  about RAID.)

-- 
Torbjörn
___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "freebsd-questions-unsubscr...@freebsd.org"

Re: TCP conection problems IBM VM -> FreeBSD

[Torbjorn Granlund]

Chuck Swiger <[EMAIL PROTECTED]> writes:

  On Mar 22, 2007, at 12:00 PM, Torbjorn Granlund wrote:
  >>   The second line should have been smtp.swox.se.smtp SYN+ACK'ing the
  >>   ISN of 27523124.  vm is sending a RST to that because the sequence
  >>   #'s don't match.  It's also odd that the set of options being
  >> listed
  >>   don't correspond at all...if you run the tcpdump for several
  >> minutes,
  >>   can you track down other SYN requests which do correspond?
  >
  > These are the ones the correspond.  They come in bursts like that.  If
  > I let it run a little longer, I get output like this:
  >
  > 19:45:56.939958 IP vm.se.lsoft.com.58679 > bang.swox.se.smtp: S
  > 678305700:678305700(0) win 8192  0,nop,nop,nop,timestamp 2317060084 0>
  > 19:45:56.940154 IP bang.swox.se.smtp > vm.se.lsoft.com.58679: S
  > 3183232720:3183232720(0) ack 678305701 win 57344  1460,nop,wscale 0,nop,nop,timestamp 24588210 2317060084>
  
  Notice the ACK from vm.se.lsoft.com is off by one, but the timestamp
  option corresponds.  Looks to be a bug with the vm machine, the bang
  machine is behaving properly per the TCP requirements.
  
Now you're confusing me.

(1) There is no ACK sent from vm.se.lsoft.com.  The only ACKs sent
are i  the other directions, as per the tcpdumps above.
(2) The ACKs are not off by one, or if you prefer, ACKs on TCP are
always off by one.  They are one higher than one would expect.

-- 
Torbjörn
___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[EMAIL PROTECTED]"

Re: TCP conection problems IBM VM -> FreeBSD

[Torbjorn Granlund]

Chuck Swiger <[EMAIL PROTECTED]> writes:

  On Mar 21, 2007, at 5:03 PM, Torbjorn Granlund wrote:
When vm attempts to make a TCP connection (e.g., on port 25) to
smtp.swox.se I see the following traffic on the router:
   
22:46:27.015389 IP vm.se.lsoft.com.47218 > smtp.swox.se.smtp: S 
27523124:27523124(0) win 8192 
22:46:27.015523 IP smtp.swox.se.smtp > vm.se.lsoft.com.47218: S 
1745147473:1745147473(0) ack 3530628660 win 57344 
22:46:27.056277 IP vm.se.lsoft.com.47218 > smtp.swox.se.smtp: R 
3530628660:3530628660(0) win 0
  
I.e., the vm box appears to dislike the SYNACK from smtp.swox.se, and
sends an RST.  One might ask if it is the fault of vm or of  
smtp.swox.se.
  
  The second line should have been smtp.swox.se.smtp SYN+ACK'ing the  
  ISN of 27523124.  vm is sending a RST to that because the sequence  
  #'s don't match.  It's also odd that the set of options being listed  
  don't correspond at all...if you run the tcpdump for several minutes,  
  can you track down other SYN requests which do correspond?

These are the ones the correspond.  They come in bursts like that.  If
I let it run a little longer, I get output like this:
  
19:45:56.939958 IP vm.se.lsoft.com.58679 > bang.swox.se.smtp: S 
678305700:678305700(0) win 8192 
19:45:56.940154 IP bang.swox.se.smtp > vm.se.lsoft.com.58679: S 
3183232720:3183232720(0) ack 678305701 win 57344 
19:45:56.974421 IP vm.se.lsoft.com.58679 > bang.swox.se.smtp: R 
678305701:678305701(0) win 0
19:45:59.939737 IP vm.se.lsoft.com.58679 > bang.swox.se.smtp: S 
678305700:678305700(0) win 8192 
19:45:59.939905 IP bang.swox.se.smtp > vm.se.lsoft.com.58679: S 
1749284606:1749284606(0) ack 678305701 win 57344 
19:45:59.978666 IP vm.se.lsoft.com.58679 > bang.swox.se.smtp: R 
678305701:678305701(0) win 0
19:46:05.940041 IP vm.se.lsoft.com.58679 > bang.swox.se.smtp: S 
678305700:678305700(0) win 8192 
19:46:05.940205 IP bang.swox.se.smtp > vm.se.lsoft.com.58679: S 
2664894402:2664894402(0) ack 678305701 win 57344 
19:46:05.977251 IP vm.se.lsoft.com.58679 > bang.swox.se.smtp: R 
678305701:678305701(0) win 0
  
The ISN's don't match here either.

  Sometimes this kind of re-writing can happen if natd or PF is  
  attempting to translate the packets, perhaps when they shouldn't if  
  both sides of your router box are using routable IPs
  
I don't run natd at all, and to get the output above from tcpdump I
had disabled pf with pfctl -d.  With pf running, it silently drops the
2nd packet.  Could that too be related to ISN's?

The outside of the fbsd 6.2 router has two addresses, one routable and
one not routable.  This is due to the default setup my ISP is
providing: Their is a little net 192.168.0.0/30 between their router
and my fbsd 6.2 router.

(I have a routable address on the interface in order to allow pf's nat
to provide a sensible return address for the nat'ed packets.)

-- 
Torbjörn
___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[EMAIL PROTECTED]"

TCP conection problems IBM VM -> FreeBSD

[Torbjorn Granlund]

I have the following setup:

IBM VM mainframe <-> [Internet] <->  fbsd 6.2 router   <-> fbsd 4.10 smtp box
"vm" "router.swox.se"   "smtp.swox.se"

When vm attempts to make a TCP connection (e.g., on port 25) to
smtp.swox.se I see the following traffic on the router:

22:46:27.015389 IP vm.se.lsoft.com.47218 > smtp.swox.se.smtp: S 
27523124:27523124(0) win 8192 
22:46:27.015523 IP smtp.swox.se.smtp > vm.se.lsoft.com.47218: S 
1745147473:1745147473(0) ack 3530628660 win 57344 
22:46:27.056277 IP vm.se.lsoft.com.47218 > smtp.swox.se.smtp: R 
3530628660:3530628660(0) win 0

I.e., the vm box appears to dislike the SYNACK from smtp.swox.se, and
sends an RST.  One might ask if it is the fault of vm or of smtp.swox.se.

If I switch on "pf", the kernel packet filter, on router.swox.se, using
pfctl -e, with an empty /etc/pf.conf, the router discards the SYNACK
from smtp.swox.se.  I realize that an empty /etc/pf.conf means that a
set of implicit rules are enabled, including some sanitize rules.

This leads me to believing that smtp.swox.se somehow sends a bad
SYNACK reply to vm.  But smtp.swox.se is perfectly capable of
accepting TCP connections from lots of machines out there, and the
router leavs the SYNACKs alone except when vm is on the receiving end.

I have stared at tcpdumps in order to try to understand what might be
wrong, unsuccessfully.  I have not found out how to make pf on
router.swox.se be so kind as to log the reason for that it throws away
the SYNACK.  And I don't have access to nor knowledge to of the vm
system to get any information on why it dislikes the SYNACK.

Making tcp connections in the other direction (smtp.swox.se -> vm)
works flawlessly.

I have tried enabling and disabling rfc1323 on smtp.swox.se, without
any change wrt this issue.

My questions:

  How can I debug this further?

  Was there a tcp bug in 4.10 that might be causing the observed
  behaviour?  (Is it fixed in 4.11?)

(I might be upgrading smtp.swox.se to a much more recent FreeBSD
version at some point, but I'd rather not hurry up with it since this
machine runs a lot of things.  Understanding this problem is important
in any case.)

-- 
Torbjörn
___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[EMAIL PROTECTED]"