hot scsi disk resize

hot scsi disk resize

[Alex Tomas]

Hello!

Here is patch to implement hot scsi resize function.
Modern storage boxes support virtualization feature and
may change logical volume size online. I think it would
be great if linux supports such abilities. Look at
this example:

root@zefir:# dmesg | tail
scsi0:A:1:0: Tagged Queuing enabled.  Depth 32
SCSI device sda: 2097152 512-byte hdwr sectors (1074 MB)
SCSI device sda: drive cache: write through
 sda: unknown partition table
Attached scsi disk sda at scsi0, channel 0, id 1, lun 0

root@zefir:# mount -treiserfs /dev/sda /mnt

root@zefir:# df -Th
Filesystem    Type    Size  Used Avail Use% Mounted on
/dev/hda1     ext3    1.9G  1.8G  138M  93% /
/dev/sda  reiserfs    1.0G   33M  991M   4% /mnt

root@zefir:# echo 'scsi rescan 0 0 1 0' >/proc/scsi/scsi 
root@zefir:# dmesg|tail -n1
SCSI device sda: 20971520 512-byte hdwr sectors (10737 MB)
root@zefir:# resize_reiserfs /dev/sda 

<-------------resize_reiserfs, 2002------------->
reiserfsprogs 3.6.4

root@zefir:# df -Th
Filesystem    Type    Size  Used Avail Use% Mounted on
/dev/hda1     ext3    1.9G  1.8G  138M  93% /
/dev/sda  reiserfs     10G   33M  9.9G   1% /mnt

I think it's really useful for for high-availability systems.
MS Windows has supported online disk resize since 2000. I Would
be happy to hear any comments and suggestions.




diff -uNr linux/drivers/scsi/hosts.h edited/drivers/scsi/hosts.h
--- linux/drivers/scsi/hosts.h	Thu Feb 20 13:18:16 2003
+++ edited/drivers/scsi/hosts.h	Mon Mar  3 18:48:16 2003
@@ -558,6 +558,7 @@
     void (*detach)(Scsi_Device *);
     int (*init_command)(Scsi_Cmnd *);     /* Used by new queueing code. 
                                            Selects command for blkdevs */
+    void (*rescan)(Scsi_Device *);
     struct device_driver scsi_driverfs_driver;
 };
 
diff -uNr linux/drivers/scsi/scsi.c edited/drivers/scsi/scsi.c
--- linux/drivers/scsi/scsi.c	Thu Feb 20 13:18:17 2003
+++ edited/drivers/scsi/scsi.c	Mon Mar  3 18:48:16 2003
@@ -1510,6 +1510,21 @@
 	up_read(&scsi_devicelist_mutex);
 }
 
+void scsi_rescan_device(struct scsi_device *sdev)
+{
+	struct Scsi_Device_Template *sdt;
+
+	down_read(&scsi_devicelist_mutex);
+	list_for_each_entry(sdt, &scsi_devicelist, list) {
+		if (!try_module_get(sdt->module))
+			continue;
+		if (*sdt->rescan)
+			(*sdt->rescan)(sdev);
+		module_put(sdt->module);
+	}
+	up_read(&scsi_devicelist_mutex);
+}
+
 int scsi_device_get(struct scsi_device *sdev)
 {
 	if (!try_module_get(sdev->host->hostt->module))
diff -uNr linux/drivers/scsi/scsi_proc.c edited/drivers/scsi/scsi_proc.c
--- linux/drivers/scsi/scsi_proc.c	Thu Feb 20 13:18:17 2003
+++ edited/drivers/scsi/scsi_proc.c	Mon Mar  3 18:48:16 2003
@@ -551,6 +551,18 @@
 		lun = simple_strtoul(p + 1, &p, 0);
 
 		err = scsi_remove_single_device(host, channel, id, lun);
+		if (err >= 0)
+			err = length;
+	} else if (!strncmp("rescan", buffer + 5, 6)) {
+		p = buffer + 12;
+
+		host = simple_strtoul(p, &p, 0);
+		channel = simple_strtoul(p + 1, &p, 0);
+		id = simple_strtoul(p + 1, &p, 0);
+		lun = simple_strtoul(p + 1, &p, 0);
+		err = scsi_rescan_single_device(host, channel, id, lun);
+		if (err >= 0)
+			err = length;
 	}
 out:
 	
diff -uNr linux/drivers/scsi/scsi_scan.c edited/drivers/scsi/scsi_scan.c
--- linux/drivers/scsi/scsi_scan.c	Thu Feb 20 13:18:17 2003
+++ edited/drivers/scsi/scsi_scan.c	Mon Mar  3 18:48:16 2003
@@ -1796,6 +1796,24 @@
 	return error;
 }
 
+int scsi_rescan_single_device(uint host, uint channel, uint id, uint lun)
+{
+	struct scsi_device *sdev;
+	struct Scsi_Host *shost;
+
+	shost = scsi_host_hn_get(host);
+	if (!shost)
+		return -ENODEV;
+	sdev = scsi_find_device(shost, channel, id, lun);
+	if (!sdev)
+		goto out;
+
+	scsi_rescan_device(sdev);	
+out:
+	scsi_host_put(shost);
+	return 0;
+}
+
 /**
  * scsi_scan_target - scan a target id, possibly including all LUNs on the
  *     target.
diff -uNr linux/drivers/scsi/sd.c edited/drivers/scsi/sd.c
--- linux/drivers/scsi/sd.c	Mon Jan 20 02:23:42 2003
+++ edited/drivers/scsi/sd.c	Mon Mar  3 19:19:25 2003
@@ -93,10 +93,12 @@
 
 static int sd_attach(struct scsi_device *);
 static void sd_detach(struct scsi_device *);
+static void sd_rescan(struct scsi_device *);
 static int sd_init_command(struct scsi_cmnd *);
 static int sd_synchronize_cache(struct scsi_disk *, int);
 static int sd_notifier(struct notifier_block *, unsigned long, void *);
-
+static void sd_read_capacity(struct scsi_disk *sdkp, char *diskname,
+		 struct scsi_request *SRpnt, unsigned char *buffer);
 static struct notifier_block sd_notifier_block = {sd_notifier, NULL, 0}; 
 
 static struct Scsi_Device_Template sd_template = {
@@ -106,6 +108,7 @@
 	.scsi_type	= TYPE_DISK,
 	.attach		= sd_attach,
 	.detach		= sd_detach,
+	.rescan		= sd_rescan,
 	.init_command	= sd_init_command,
 	.scsi_driverfs_driver = {
 		.name   = "sd",
@@ -627,6 +630,38 @@
 not_present:
 	set_media_not_present(sdkp);
 	return 1;
+}
+
+static void sd_rescan(struct scsi_device * sdp)
+{
+	unsigned char *buffer;
+	struct scsi_disk *sdkp = sd_find_by_sdev(sdp);
+	struct gendisk *gd;
+	struct scsi_request *SRpnt;
+
+	if (!sdkp || sdp->online == FALSE || !sdkp->media_present)
+		return;
+		
+	gd = sdkp->disk;
+	
+	SCSI_LOG_HLQUEUE(3, printk("sd_rescan: disk=%s\n", gd->disk_name));
+	
+	SRpnt = scsi_allocate_request(sdp);
+	if (!SRpnt) {
+		printk(KERN_WARNING "(sd_rescan:) Request allocation "
+		       "failure.\n");
+		return;
+	}
+
+	if (sdkp->device->host->unchecked_isa_dma)
+		buffer = kmalloc(512, GFP_DMA);
+	else
+		buffer = kmalloc(512, GFP_KERNEL);
+
+    	sd_read_capacity(sdkp, gd->disk_name, SRpnt, buffer);
+	set_capacity(gd, sdkp->capacity);	
+	scsi_release_request(SRpnt);
+	kfree(buffer);
 }
 
 static int sd_revalidate_disk(struct gendisk *disk)
diff -uNr linux/fs/block_dev.c edited/fs/block_dev.c
--- linux/fs/block_dev.c	Mon Jan 20 02:23:49 2003
+++ edited/fs/block_dev.c	Mon Mar  3 18:49:47 2003
@@ -623,6 +623,8 @@
 			up(&whole->bd_sem);
 		}
 	} else {
+		if (!part)
+			bd_set_size(bdev,(loff_t)get_capacity(disk)<<9);
 		put_disk(disk);
 		module_put(owner);
 		if (bdev->bd_contains == bdev) {

Re: hot scsi disk resize

[Christoph Hellwig]

On Mon, Mar 03, 2003 at 08:21:24PM +0300, Alex Tomas wrote:
> +		if (err >= 0)
> +			err = length;
> +	} else if (!strncmp("rescan", buffer + 5, 6)) {
> +		p = buffer + 12;
> +
> +		host = simple_strtoul(p, &p, 0);
> +		channel = simple_strtoul(p + 1, &p, 0);
> +		id = simple_strtoul(p + 1, &p, 0);
> +		lun = simple_strtoul(p + 1, &p, 0);
> +		err = scsi_rescan_single_device(host, channel, id, lun);
> +		if (err >= 0)
> +			err = length;

Please don't add more procfs option.  A resize attribute for the
sysfs node of the scsi_device is the better choice in my opinion.

> +		buffer = kmalloc(512, GFP_DMA);
> +	else
> +		buffer = kmalloc(512, GFP_KERNEL);
> +
> +    	sd_read_capacity(sdkp, gd->disk_name, SRpnt, buffer);
> +	set_capacity(gd, sdkp->capacity);	
> +	scsi_release_request(SRpnt);
> +	kfree(buffer);

Don't you need some kind of serialization here?

> --- linux/fs/block_dev.c	Mon Jan 20 02:23:49 2003
> +++ edited/fs/block_dev.c	Mon Mar  3 18:49:47 2003
> @@ -623,6 +623,8 @@
>  			up(&whole->bd_sem);
>  		}
>  	} else {
> +		if (!part)
> +			bd_set_size(bdev,(loff_t)get_capacity(disk)<<9);

This needs more explanation.

Re: hot scsi disk resize

[Alex Tomas]

Hi!

>>>>> Christoph Hellwig (CH) writes:

 CH> On Mon, Mar 03, 2003 at 08:21:24PM +0300, Alex Tomas wrote:
 >> + if (err >= 0) + err = length; + } else if (!strncmp("rescan",
 >> buffer + 5, 6)) { + p = buffer + 12; + + host = simple_strtoul(p,
 >> &p, 0); + channel = simple_strtoul(p + 1, &p, 0); + id =
 >> simple_strtoul(p + 1, &p, 0); + lun = simple_strtoul(p + 1, &p,
 >> 0); + err = scsi_rescan_single_device(host, channel, id, lun); +
 >> if (err >= 0) + err = length;

 CH> Please don't add more procfs option.  A resize attribute for the
 CH> sysfs node of the scsi_device is the better choice in my opinion.

I like your idea.

 >> + buffer = kmalloc(512, GFP_DMA); + else + buffer = kmalloc(512,
 >> GFP_KERNEL); + + sd_read_capacity(sdkp, gd->disk_name, SRpnt,
 >> buffer); + set_capacity(gd, sdkp->capacity); +
 >> scsi_release_request(SRpnt); + kfree(buffer);

 CH> Don't you need some kind of serialization here?

Well. I think all changes sd_read_capacity() and set_capacity() make
need not to be protected. Am I right?

 >> --- linux/fs/block_dev.c Mon Jan 20 02:23:49 2003 +++
 >> edited/fs/block_dev.c Mon Mar 3 18:49:47 2003 @@ -623,6 +623,8 @@
 >> up(&whole->bd_sem); } } else { + if (!part) +
 >> bd_set_size(bdev,(loff_t)get_capacity(disk)<<9);

 CH> This needs more explanation.

This is required because open("/dev/sd*",) should see new size of
disk being used. James Bottomley suggested to introduce new ioctl, but
I hope we may implmenent the feature w/o touching userspace tools.

with best regards, Alex

Re: hot scsi disk resize

[alexey]

Hi!

Here is new version of the patch. All procfs-related stuff has been removed.
One may rescan device size writing something to /sysfs/.../<scsi device>/rescan:

root@zefir:~# echo 1 >/sysfs/bus/scsi/devices/0\:0\:1\:0/rescan 
root@zefir:~# dmesg
scsi0:A:1:0: Tagged Queuing enabled.  Depth 64
scsi: host 0 channel 0 id 1 lun16384 has a LUN larger than allowed by the host adapter
SCSI device sda: 2097152 512-byte hdwr sectors (1074 MB)
SCSI device sda: drive cache: write through
 sda: unknown partition table
Attached scsi disk sda at scsi0, channel 0, id 1, lun 0
SCSI device sda: 125829120 512-byte hdwr sectors (64425 MB)
root@zefir:~# 



diff -uNr linux/drivers/scsi/hosts.h edited/drivers/scsi/hosts.h
--- linux/drivers/scsi/hosts.h	Thu Feb 20 13:18:16 2003
+++ edited/drivers/scsi/hosts.h	Mon Mar  3 18:48:16 2003
@@ -558,6 +558,7 @@
     void (*detach)(Scsi_Device *);
     int (*init_command)(Scsi_Cmnd *);     /* Used by new queueing code. 
                                            Selects command for blkdevs */
+    void (*rescan)(Scsi_Device *);
     struct device_driver scsi_driverfs_driver;
 };
 
diff -uNr linux/drivers/scsi/scsi.c edited/drivers/scsi/scsi.c
--- linux/drivers/scsi/scsi.c	Thu Feb 20 13:18:17 2003
+++ edited/drivers/scsi/scsi.c	Mon Mar  3 18:48:16 2003
@@ -1510,6 +1510,21 @@
 	up_read(&scsi_devicelist_mutex);
 }
 
+void scsi_rescan_device(struct scsi_device *sdev)
+{
+	struct Scsi_Device_Template *sdt;
+
+	down_read(&scsi_devicelist_mutex);
+	list_for_each_entry(sdt, &scsi_devicelist, list) {
+		if (!try_module_get(sdt->module))
+			continue;
+		if (*sdt->rescan)
+			(*sdt->rescan)(sdev);
+		module_put(sdt->module);
+	}
+	up_read(&scsi_devicelist_mutex);
+}
+
 int scsi_device_get(struct scsi_device *sdev)
 {
 	if (!try_module_get(sdev->host->hostt->module))
diff -uNr linux/drivers/scsi/scsi_sysfs.c edited/drivers/scsi/scsi_sysfs.c
--- linux/drivers/scsi/scsi_sysfs.c	Thu Feb 20 13:18:17 2003
+++ edited/drivers/scsi/scsi_sysfs.c	Tue Mar  4 09:15:42 2003
@@ -219,6 +219,25 @@
 sdev_rd_attr (rev, "%.4s\n");
 sdev_rw_attr_bit (online);
 
+static ssize_t
+show_rescan_field (struct device *dev, char *buf)
+{
+	return 0; 
+}
+
+static ssize_t
+store_rescan_field (struct device *dev, const char *buf, size_t count) 
+{
+	int ret = ENODEV;
+	struct scsi_device *sdev;
+	sdev = to_scsi_device(dev);
+	if (sdev)
+		ret = scsi_rescan_device(sdev);
+	return ret;
+}
+
+static DEVICE_ATTR(rescan, S_IRUGO | S_IWUSR, show_rescan_field, store_rescan_field)
+
 static struct device_attribute * const sdev_attrs[] = {
 	&dev_attr_device_blocked,
 	&dev_attr_queue_depth,
@@ -229,6 +248,7 @@
 	&dev_attr_model,
 	&dev_attr_rev,
 	&dev_attr_online,
+	&dev_attr_rescan,
 };
 
 /**
diff -uNr linux/drivers/scsi/sd.c edited/drivers/scsi/sd.c
--- linux/drivers/scsi/sd.c	Mon Jan 20 02:23:42 2003
+++ edited/drivers/scsi/sd.c	Mon Mar  3 19:19:25 2003
@@ -93,10 +93,12 @@
 
 static int sd_attach(struct scsi_device *);
 static void sd_detach(struct scsi_device *);
+static void sd_rescan(struct scsi_device *);
 static int sd_init_command(struct scsi_cmnd *);
 static int sd_synchronize_cache(struct scsi_disk *, int);
 static int sd_notifier(struct notifier_block *, unsigned long, void *);
-
+static void sd_read_capacity(struct scsi_disk *sdkp, char *diskname,
+		 struct scsi_request *SRpnt, unsigned char *buffer);
 static struct notifier_block sd_notifier_block = {sd_notifier, NULL, 0}; 
 
 static struct Scsi_Device_Template sd_template = {
@@ -106,6 +108,7 @@
 	.scsi_type	= TYPE_DISK,
 	.attach		= sd_attach,
 	.detach		= sd_detach,
+	.rescan		= sd_rescan,
 	.init_command	= sd_init_command,
 	.scsi_driverfs_driver = {
 		.name   = "sd",
@@ -627,6 +630,38 @@
 not_present:
 	set_media_not_present(sdkp);
 	return 1;
+}
+
+static void sd_rescan(struct scsi_device * sdp)
+{
+	unsigned char *buffer;
+	struct scsi_disk *sdkp = sd_find_by_sdev(sdp);
+	struct gendisk *gd;
+	struct scsi_request *SRpnt;
+
+	if (!sdkp || sdp->online == FALSE || !sdkp->media_present)
+		return;
+		
+	gd = sdkp->disk;
+	
+	SCSI_LOG_HLQUEUE(3, printk("sd_rescan: disk=%s\n", gd->disk_name));
+	
+	SRpnt = scsi_allocate_request(sdp);
+	if (!SRpnt) {
+		printk(KERN_WARNING "(sd_rescan:) Request allocation "
+		       "failure.\n");
+		return;
+	}
+
+	if (sdkp->device->host->unchecked_isa_dma)
+		buffer = kmalloc(512, GFP_DMA);
+	else
+		buffer = kmalloc(512, GFP_KERNEL);
+
+    	sd_read_capacity(sdkp, gd->disk_name, SRpnt, buffer);
+	set_capacity(gd, sdkp->capacity);	
+	scsi_release_request(SRpnt);
+	kfree(buffer);
 }
 
 static int sd_revalidate_disk(struct gendisk *disk)
diff -uNr linux/fs/block_dev.c edited/fs/block_dev.c
--- linux/fs/block_dev.c	Mon Jan 20 02:23:49 2003
+++ edited/fs/block_dev.c	Mon Mar  3 18:49:47 2003
@@ -546,7 +546,7 @@
 	return res;
 }
 
-static void bd_set_size(struct block_device *bdev, loff_t size)
+void bd_set_size(struct block_device *bdev, loff_t size)
 {
 	unsigned bsize = bdev_hardsect_size(bdev);
 	bdev->bd_inode->i_size = size;
@@ -623,6 +623,8 @@
 			up(&whole->bd_sem);
 		}
 	} else {
+		if (!part)
+			bd_set_size(bdev,(loff_t)get_capacity(disk)<<9);
 		put_disk(disk);
 		module_put(owner);
 		if (bdev->bd_contains == bdev) {

Re: hot scsi disk resize

[James Bottomley]

On Tue, 2003-03-04 at 01:00, alexey@technomagesinc.com wrote:
> Here is new version of the patch. All procfs-related stuff has been removed.
> One may rescan device size writing something to /sysfs/.../<scsi device>/rescan:

I applied all of this, except this piece:

> diff -uNr linux/fs/block_dev.c edited/fs/block_dev.c
> --- linux/fs/block_dev.c	Mon Jan 20 02:23:49 2003
> +++ edited/fs/block_dev.c	Mon Mar  3 18:49:47 2003
> @@ -623,6 +623,8 @@
>  			up(&whole->bd_sem);
>  		}
>  	} else {
> +		if (!part)
> +			bd_set_size(bdev,(loff_t)get_capacity(disk)<<9);
>  		put_disk(disk);
>  		module_put(owner);
>  		if (bdev->bd_contains == bdev) {
> 
> 

Because adding this in will break the initial ramdisk.  Thus, although
the SCSI layer does resize, the block layer may not.

I've copied Al Viro because he's more familiar with the block layer
device code than I and may have been planning to add this feature.  Al,
can you tell us what needs to happen to allow for hot resizing a SCSI
disc?

Thanks,

James

Re: hot scsi disk resize

[Christoph Hellwig]

On Mon, Mar 17, 2003 at 10:21:41AM -0600, James Bottomley wrote:
> > --- linux/fs/block_dev.c	Mon Jan 20 02:23:49 2003
> > +++ edited/fs/block_dev.c	Mon Mar  3 18:49:47 2003
> > @@ -623,6 +623,8 @@
> >  			up(&whole->bd_sem);
> >  		}
> >  	} else {
> > +		if (!part)
> > +			bd_set_size(bdev,(loff_t)get_capacity(disk)<<9);
> >  		put_disk(disk);
> >  		module_put(owner);
> >  		if (bdev->bd_contains == bdev) {
> > 
> > 
> 
> Because adding this in will break the initial ramdisk.  Thus, although
> the SCSI layer does resize, the block layer may not.

What about moving the size setting into ->open?

Re: hot scsi disk resize

[Fabien Salvi]

Alex Tomas a écrit:
> Hello!
> 
> Here is patch to implement hot scsi resize function.
> Modern storage boxes support virtualization feature and
> may change logical volume size online. I think it would
> be great if linux supports such abilities. Look at
> this example:
> 
> root@zefir:# dmesg | tail
> scsi0:A:1:0: Tagged Queuing enabled.  Depth 32
> SCSI device sda: 2097152 512-byte hdwr sectors (1074 MB)
> SCSI device sda: drive cache: write through
>  sda: unknown partition table
> Attached scsi disk sda at scsi0, channel 0, id 1, lun 0
> 
> root@zefir:# mount -treiserfs /dev/sda /mnt
> 
> root@zefir:# df -Th
> Filesystem    Type    Size  Used Avail Use% Mounted on
> /dev/hda1     ext3    1.9G  1.8G  138M  93% /
> /dev/sda  reiserfs    1.0G   33M  991M   4% /mnt
> 
> root@zefir:# echo 'scsi rescan 0 0 1 0' >/proc/scsi/scsi 
> root@zefir:# dmesg|tail -n1
> SCSI device sda: 20971520 512-byte hdwr sectors (10737 MB)
> root@zefir:# resize_reiserfs /dev/sda 
> 
> <-------------resize_reiserfs, 2002------------->
> reiserfsprogs 3.6.4
> 
> root@zefir:# df -Th
> Filesystem    Type    Size  Used Avail Use% Mounted on
> /dev/hda1     ext3    1.9G  1.8G  138M  93% /
> /dev/sda  reiserfs     10G   33M  9.9G   1% /mnt

Well, I also think it's a very important and needed feature for HA systems.
I'm very interested in such things...

But, I thought resize_reiserfs only resize *unmounted* filesystem :
http://www.reiserfs.org/resize_reiserfs.html

Will it work under IO load ?

Reiserfs has a way to do it dynamically while using it, with mount option :
http://www.reiserfs.org/mount-options.html

I've tested this without a lot of success (I encounter a lot of problems).


I'm afraid that even if scsi subsystem support the resize option, 
FileSystems won't support it correctly...
Maybe with LVM ?

Did someone test it ?

-- 
Fabien SALVI      Centre de Ressources Informatiques
                   Archamps, France -- http://www.cri74.org
                   PingOO GNU/linux distribution : http://www.pingoo.org

-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: hot scsi disk resize

[alexey]

>>>>> Fabien Salvi (FS) writes:

 FS> Well, I also think it's a very important and needed feature for
 FS> HA systems.  I'm very interested in such things...

 FS> But, I thought resize_reiserfs only resize *unmounted* filesystem
 FS> : http://www.reiserfs.org/resize_reiserfs.html

citate from this URL: "The resize_reiserfs program allows to grow a
reiserfs on-line if there is a free space on block device."

 FS> Will it work under IO load ?

I tested it several times under different load (read-only, write-only,
read-write, single-thread and multi-thread tests). I think reiserfs
team may give comments.


 FS> Reiserfs has a way to do it dynamically while using it, with
 FS> mount option : http://www.reiserfs.org/mount-options.html

 FS> I've tested this without a lot of success (I encounter a lot of
 FS> problems).

what kind of problem have you seen? what versions of kernel/reiserfs/tools?

 FS> I'm afraid that even if scsi subsystem support the resize option,
 FS> FileSystems won't support it correctly...  Maybe with LVM ?

 FS> Did someone test it ?

>From filesystem point of view, LVM and resizable SCSI HDD are the same
think - just block device which may grow.

I think this patch is just one more step in the right direction.

Re: hot scsi disk resize

[Fabien Salvi]

alexey@technomagesinc.com a écrit:
>>>>>>Fabien Salvi (FS) writes:
>>>>>
> 
>  FS> Well, I also think it's a very important and needed feature for
>  FS> HA systems.  I'm very interested in such things...
> 
>  FS> But, I thought resize_reiserfs only resize *unmounted* filesystem
>  FS> : http://www.reiserfs.org/resize_reiserfs.html
> 
> citate from this URL: "The resize_reiserfs program allows to grow a
> reiserfs on-line if there is a free space on block device."

Yes, you're right !
I haven't seen this...
It's a bit strange, it contradicts what is said on the first line :)


>  FS> Will it work under IO load ?
> 
> I tested it several times under different load (read-only, write-only,
> read-write, single-thread and multi-thread tests). I think reiserfs
> team may give comments.


>  FS> Reiserfs has a way to do it dynamically while using it, with
>  FS> mount option : http://www.reiserfs.org/mount-options.html
> 
>  FS> I've tested this without a lot of success (I encounter a lot of
>  FS> problems).
> 
> what kind of problem have you seen? what versions of kernel/reiserfs/tools?

Well, I made some tests one year ago, I will try this again, as you say 
it seems to work well, it's interested...
I will let you know my new results.

>  FS> I'm afraid that even if scsi subsystem support the resize option,
>  FS> FileSystems won't support it correctly...  Maybe with LVM ?
> 
>  FS> Did someone test it ?
> 
>>From filesystem point of view, LVM and resizable SCSI HDD are the same
> think - just block device which may grow.
> 
> I think this patch is just one more step in the right direction.

Yes, surely !

-- 
Fabien SALVI      Centre de Ressources Informatiques
                   Archamps, France -- http://www.cri74.org
                   PingOO GNU/linux distribution : http://www.pingoo.org

-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html