On Mon, 2006-07-31 at 18:51 -0400, Greg Freemyer wrote:

On 7/31/06, Roger Oberholtzer wrote:

...

Of course hot swapping SATA disks is not currently supported in Linux. As such, udevmonitor reports nothing when the disks are inserted or removed on a system that does support this at the hardware level. There is work in libata, but nothing functioning for use in a production system.

Not sure how "production" you need it to be.

Mere mortals will need to use it. So it can't require fiddling. And it should be reliable. I wonder which SATA controllers are supported. I have an Intel 82801GB/GR/GH (ICH7 Family).

The 2.6.17 kernel has been out for a couple months I think.

Interesting. I'm using 2.6.16.13-4-smp on SUSE 10.1. Close but no cigar. Seems there is no kernel-of-the-day for 2.6.17. Unless it is somewhere other than in http://ftp.suse.com/pub/projects/kernel/kotd/ or I am looking in the wrong directory. I don't relish installing a new vanilla kernel as it messes up all the SUSE kernel packaging. Any other options? Maybe I can just update the libata.ko module. Probably not, as that would potentially be too easy.

There is a stable libata patch to add hotswap capability to it. It still has a nasty experimental warning I think in the readme, but in the several months it has been out, there really have been very few issues and none I recall caused data loss. I would definately consider it.

I wonder if this patch could be applied to the 2.6.16 kernel used in SUSE 10.0. I have looked at the libata patch directory and it seems it is a complete patch against the kernel. And that directory is not clear what is the start point to patch. Maybe I should just download a 2.6.17 kernel, grab libata, and see what happens.

The patch is also in 2.6.18-rc3, but that I too would stay away from. Not because I don't trust the patch, it is the rest of 2.6.18-rc3 I would worry about.

Quoting the July 10 announcement e-mail: ===== Updated version available.

* Link resume handling in the previous version was broken causing libata to ignore hotplug event after a link has been hot-unplugged. Fixed.

* A few other hotplug related problems are fixed.

I expect this version to have well-behaving PMP and hotplug support. If anything seems weird, please report.

More info can be found at the following URL.

http://home-tj.org/wiki/index.php/Libata-tj-stable

Updated patches against v2.6.17.4 are at the following URL.

http://home-tj.org/files/libata-tj-stable/libata-tj-2.6.17.4-20060710.tar.bz...

...

So, what would be the next best thing? I assume I would need to do a script that handles un-mounting the disks, and then a rescan of the devices. And then a re-mount.. I remember doing this way back with SCSI devices. But perhaps there is a best way to scan for SATA devices?

The tricky bit is that I have 4 SATA disks that will sort of work in sets. And the user will probably want to pop in one in the set when looking for something.

I do seem to be rambling a bit, but it is because I can't decide the best approach.

-- Roger Oberholtzer

OPQ Systems AB Ramböll Sverige AB Kapellgränd 7 P.O. Box 4205 SE-102 65 Stockholm, Sweden

Tel: Int +46 8-615 60 20 Fax: Int +46 8-31 42 23

-- Check the headers for your unsubscription address For additional commands send e-mail to suse-linux-e-help@suse.com Also check the archives at http://lists.suse.com Please read the FAQs: suse-linux-e-faq@suse.com

-- Greg Freemyer The Norcross Group Forensics for the 21st Century

-- Roger Oberholtzer OPQ Systems AB Ramböll Sverige AB Kapellgränd 7 P.O. Box 4205 SE-102 65 Stockholm, Sweden Tel: Int +46 8-615 60 20 Fax: Int +46 8-31 42 23

On Wed, 2006-08-02 at 12:19 -0400, Greg Freemyer wrote:

On 8/1/06, Roger Oberholtzer wrote:

...

Seems there is no kernel-of-the-day for 2.6.17. Unless it is somewhere other than in http://ftp.suse.com/pub/projects/kernel/kotd/ or I am looking in the wrong directory.

I was just thinking about your above comment that the newest susefied kernel was 2.6.16.

Seemed wrong, so I went looking and found both 2.6.17 & 2.6.18. I think both have SuSE specific patches applied. Not sure if the 2.6.17 will still accept the big libata patch or not. Or maybe SUSE has already applied it?

ftp://ftp.gwdg.de/pub/linux/suse/ftp.suse.com/projects/kernel/kotd/i386/HEAD/

As well as where I mentioned. However, check the dates: late on Aug 1, which is after my post! Still, I will have a check. I think this is in preparation for 10.2. Thanks for the pointer. -- Roger Oberholtzer

teck wrote:

hello all...

i set up my machine with the second nic and enabled masq all works great except! internet users can no longer ftp. the external card should accept ftp connections and allow passiv ports... Its fine it it doesnt goto my lan (2nd card) but i would like to re enable the passive ftp function... i looked thro yast and see only a ref to allow tftpd on the external... i know there is something i have overlooked...... pop http https and everything else is working just fine.....

Passive FTP should work through NAT and active mode normally doesn't. Give it a try with a browser or an FTP client that can use passive mode. The Linux FTP client defaults to passive, but the Windows client is active mode only. If the firewall doesn't explicitly list FTP, you can always specify the port number.

teck wrote:

----- Original Message ----- From: "James Knott" To: Sent: Wednesday, August 02, 2006 9:41 PM Subject: Re: [SLE] SPAM: Passive ftp through firewall?

...

teck wrote:

...

hello all...

i set up my machine with the second nic and enabled masq all works great except! internet users can no longer ftp. the external card should accept ftp connections and allow passiv ports... Its fine it it doesnt goto my lan (2nd card) but i would like to re enable the passive ftp function... i looked thro yast and see only a ref to allow tftpd on the external... i know there is something i have overlooked...... pop http https and everything else is working just fine.....

Passive FTP should work through NAT and active mode normally doesn't. Give it a try with a browser or an FTP client that can use passive mode. The Linux FTP client defaults to passive, but the Windows client is active mode only. If the firewall doesn't explicitly list FTP, you can always specify the port number.

yup i had someone try it from "the outside" with cute ftp it works fine... but my previously working unlimitedftp java script is what has quit....

What happens if you use the Linux command line FTP?

On Thu, 2006-08-03 at 05:50 -0400, Damon Register wrote:

James Knott wrote:

...

Passive FTP should work through NAT and active mode normally doesn't. Give it a try with a browser or an FTP client that can use passive mode. That's interesting. I seem to have just the opposite experience. I have to select active mode to get it to work. I am not sure though that I am doing what the original poster is doing. Are you trying to FTP in from the outside or out from the inside?

I haven't seen the beginning of this. Is this a system acting as a router i.e. 2 nics? If so are the correct netfilter modules in place? ip_conntrack_ftp ip_nat_ftp -- Dave Cotton

On Mon, 2006-07-31 at 18:51 -0400, Greg Freemyer wrote:

On 7/31/06, Roger Oberholtzer wrote:

...

Of course hot swapping SATA disks is not currently supported in Linux. As such, udevmonitor reports nothing when the disks are inserted or removed on a system that does support this at the hardware level. There is work in libata, but nothing functioning for use in a production system.

Not sure how "production" you need it to be.

The 2.6.17 kernel has been out for a couple months I think.

Wonder if it works with intel chipsets (ahci). I installed 2.6.17.7, and still did not see udev activity. Maybe it is only working for certain chip sets. Not much in the Docs in the kernel source. Oh well. Back to plan A: a script. Any suggestions on the best way to get the system to rescan for sata drives? I used to have something that used /proc/scsi. But isn't that going away?

There is a stable libata patch to add hotswap capability to it. It still has a nasty experimental warning I think in the readme, but in the several months it has been out, there really have been very few issues and none I recall caused data loss. I would definately consider it.

The patch is also in 2.6.18-rc3, but that I too would stay away from. Not because I don't trust the patch, it is the rest of 2.6.18-rc3 I would worry about.

Quoting the July 10 announcement e-mail: ===== Updated version available.

* Link resume handling in the previous version was broken causing libata to ignore hotplug event after a link has been hot-unplugged. Fixed.

* A few other hotplug related problems are fixed.

I expect this version to have well-behaving PMP and hotplug support. If anything seems weird, please report.

More info can be found at the following URL.

http://home-tj.org/wiki/index.php/Libata-tj-stable

Updated patches against v2.6.17.4 are at the following URL.

http://home-tj.org/files/libata-tj-stable/libata-tj-2.6.17.4-20060710.tar.bz...

...

So, what would be the next best thing? I assume I would need to do a script that handles un-mounting the disks, and then a rescan of the devices. And then a re-mount.. I remember doing this way back with SCSI devices. But perhaps there is a best way to scan for SATA devices?

The tricky bit is that I have 4 SATA disks that will sort of work in sets. And the user will probably want to pop in one in the set when looking for something.

I do seem to be rambling a bit, but it is because I can't decide the best approach.

-- Roger Oberholtzer

OPQ Systems AB Ramböll Sverige AB Kapellgränd 7 P.O. Box 4205 SE-102 65 Stockholm, Sweden

Tel: Int +46 8-615 60 20 Fax: Int +46 8-31 42 23

-- Check the headers for your unsubscription address For additional commands send e-mail to suse-linux-e-help@suse.com Also check the archives at http://lists.suse.com Please read the FAQs: suse-linux-e-faq@suse.com

-- Greg Freemyer The Norcross Group Forensics for the 21st Century

-- Roger Oberholtzer OPQ Systems AB Ramböll Sverige AB Kapellgränd 7 P.O. Box 4205 SE-102 65 Stockholm, Sweden Tel: Int +46 8-615 60 20 Fax: Int +46 8-31 42 23

On Tue, 2006-08-01 at 16:33 -0400, Greg Freemyer wrote:

On 8/1/06, Roger Oberholtzer wrote:

...

On Mon, 2006-07-31 at 18:51 -0400, Greg Freemyer wrote:

...

On 7/31/06, Roger Oberholtzer wrote:

...

Of course hot swapping SATA disks is not currently supported in Linux. As such, udevmonitor reports nothing when the disks are inserted or removed on a system that does support this at the hardware level. There is work in libata, but nothing functioning for use in a production system.

Not sure how "production" you need it to be.

The 2.6.17 kernel has been out for a couple months I think.

Wonder if it works with intel chipsets (ahci). I installed 2.6.17.7, and still did not see udev activity. Maybe it is only working for certain chip sets. Not much in the Docs in the kernel source.

Oh well. Back to plan A: a script.

Any suggestions on the best way to get the system to rescan for sata drives? I used to have something that used /proc/scsi. But isn't that going away?

Sorry, not really any ideas for the script, but for 2.6.17.7 you have to have both the vanilla kernel from kernel.org and you have to apply the patch from

http://home-tj.org/files/libata-tj-stable/libata-tj-2.6.17.4-20060710.tar.bz...

I have not tried the SUSE 2.6.17 kernel yet. But I have sorted out the hot swap of the sata disks using a vanilla kernel with this patch. I needed to put my sata controller in AHCI mode. At first I did not see this in the BIOS. Then, I did. So, now I can put in a disk and it mounts. I have 4 hot swap bays and it seems that it keeps track of which drive bay it is. All is good. Here is the question: udev handles mounting the disks via the rule in /etc/udev/rules.d/85-mount-fstab.rules (SUSE 10.0). Or at least I am pretty sure this is doing this. How can I get this to unmount the disks as well? I know all the issues, but these are mounted read only so I am not worried about disk corruption. I just want the disk to be unmounted when it is removed. Some counterpart to the RUN+="mount.sh" statement in the rules file. Any ideas?

Greg -- Greg Freemyer The Norcross Group Forensics for the 21st Century

-- Roger Oberholtzer OPQ Systems AB Ramböll Sverige AB Kapellgränd 7 P.O. Box 4205 SE-102 65 Stockholm, Sweden Tel: Int +46 8-615 60 20 Fax: Int +46 8-31 42 23