Jump to content


Photo
- - - - -

Leopard?


  • Please log in to reply
34 replies to this topic

#1 wurm

wurm

    Newbie

  • Members
  • Pip
  • 3 posts

Posted 30 October 2007 - 08:25 AM

I'm wondering if any one has gotten this to work with Leopard yet? The application runs and I can get it to log into iSCSI targets, I can see the session on the Sessions page however nothing ever shows up in Disk Utility to allow me to format/mount etc.

Am I missing a step or does this product not work with leopard yet? I don't have a Tiger machine around any more to test with or I would.

My ultimate goal is that hopefully the iSCSI disk will work with Time Machine, maybe someone already has experience with this.

Thanks!

#2 Cincinnatus

Cincinnatus

    Newbie

  • Members
  • Pip
  • 1 posts

Posted 30 October 2007 - 12:58 PM

I am also curious to know if the globalSAN Initiator supports Leopard, for the same reason as wurm: Time Machine. I have it working on Tiger, connecting to an iSCSI Enterprise Target LUN, formatted as HFS+, which works very well for remote storage.

#3 wurm

wurm

    Newbie

  • Members
  • Pip
  • 3 posts

Posted 30 October 2007 - 02:42 PM

I'm wondering if any one has gotten this to work with Leopard yet? The application runs and I can get it to log into iSCSI targets, I can see the session on the Sessions page however nothing ever shows up in Disk Utility to allow me to format/mount etc.

Am I missing a step or does this product not work with leopard yet? I don't have a Tiger machine around any more to test with or I would.

My ultimate goal is that hopefully the iSCSI disk will work with Time Machine, maybe someone already has experience with this.

Thanks!



I'd like to follow this up. The problem I'm seeing is that I can log into the target but I don't see any devices, this seems to be a problem with my Target (OpenFiler) and not anything to do with globalSAN's product. I get the same exactly results using a Windows Server 2003 machine with Microsoft's iSCSI initiator.

Once I get my issue sorted out I will post back to this topic if it works with Leopard and TimeMachine.

#4 wurm

wurm

    Newbie

  • Members
  • Pip
  • 3 posts

Posted 30 October 2007 - 06:55 PM

Final Followup:
I resolved my iSCSI issue and was able to get everything working under Leopard. The iSCSI initiator works exactly as I would expect and mounts the disk. I was able to format it HFS+ and everything looks fine... Now for the best news of it all, the drive shows up as a valid time machine target within the Time Machine preferences!

I was able to point time machine at it and it has started my backup. This is great news for those who were hoping to do network backups with time machine.

For reference my specific hardware setup is:
Core Duo Mac Mini 80gb HDD, OSX 10.5 using the latest globalSAN iSCSI Initiator
For an iSCSI target I am using Openfiler 2.2
:D

#5 example42

example42

    Newbie

  • Members
  • Pip
  • 1 posts

Posted 02 November 2007 - 03:28 PM

Just to chime in and give a second success story...

Last night I set up a spare server as an iSCSI target using the Openfiler 2.2 Linux distro. I then loaded the iSCSI initiator on my 10.4 MacBook. It connected up to the target, no problem. I backed up all my data and proceeded to then wipe the MacBook and install 10.5. I loaded up the initiator on the new install of 10.5 and it connected to the Openfiler iSCSI target and worked great. I moved everything over, wiped the iSCSI volume, and am now using it as a Time Machine drive just to give it a test spin.

Summary: No problem in 10.5 Works great!

#6 hpsan

hpsan

    Newbie

  • Members
  • Pip
  • 5 posts

Posted 03 November 2007 - 05:41 PM

How did you fix your problem?

I have Enterprise iscsi 0.4.15 installed, and cannot get a device to mount, either with leopard or tiger. For the person that is using this with Enterprise Iscsi, I'd love to know how you have things configured!!!!

I really want to use this as a time machine backup setup!


Final Followup:
I resolved my iSCSI issue and was able to get everything working under Leopard. The iSCSI initiator works exactly as I would expect and mounts the disk. I was able to format it HFS+ and everything looks fine... Now for the best news of it all, the drive shows up as a valid time machine target within the Time Machine preferences!

I was able to point time machine at it and it has started my backup. This is great news for those who were hoping to do network backups with time machine.

For reference my specific hardware setup is:
Core Duo Mac Mini 80gb HDD, OSX 10.5 using the latest globalSAN iSCSI Initiator
For an iSCSI target I am using Openfiler 2.2
:D



#7 hpsan

hpsan

    Newbie

  • Members
  • Pip
  • 5 posts

Posted 03 November 2007 - 09:55 PM

I downloaded the latest SVN and kinda got it to work. I can now log into a LUN and see it in Disk Utility! Only the first lun worked for some reason. More to look into....now backing up to time machine...

#8 Christoph

Christoph

    Newbie

  • Members
  • Pip
  • 1 posts

Posted 13 November 2007 - 09:20 AM

Hi Guys,

I'm using iSCSI Enterprise Target 0.4.15 on a ubuntu 7.04 machine.

I set up a target with 100G just to test timemachine on my iMac. Target was found, connected, partitioned, formatted, introduced to timemachine without any problems. The initial timemachine transfer started. As I was connected to my network using WLAN, the transfer speed was quite low, so I decided to go to bed.

This morning I found my Mac hanging with black screen. So I rebooted.

After the reboot, I still could see the Target and connect to it, but the partition from yesterday was missing.

Anybody got any idea what has gone wrong?

thanx in advance

Christoph


For the ones who have problems connecting to the targets:

Do not connect to a target until the wheel has stopped turning, even if the desired target is shown.

#9 al4

al4

    Newbie

  • Members
  • Pip
  • 2 posts

Posted 04 December 2007 - 10:32 AM

I downloaded the latest SVN and kinda got it to work. I can now log into a LUN and see it in Disk Utility! Only the first lun worked for some reason. More to look into....now backing up to time machine...

Where did you get the latest SVN?

#10 Wicher Minnaard

Wicher Minnaard

    Newbie

  • Members
  • Pip
  • 1 posts

Posted 21 December 2007 - 06:01 AM

I'm wondering if any one has gotten this to work with Leopard yet?

I've just got it going with the iSCSI Enterprise Target. The iSCSI volume is a dm-crypt-mapped loop-mounted disk image on an USB disk - how's that for flexibility!

Relevant ietd.conf-snippet:
Target iqn.2007-12.tld.domain.subdomain:timemachine
IncomingUser johndoe veryverysecret
OutgoingUser johndoe veryverysecret
Lun 0 Path=/dev/mapper/timemachine,Type=blockio
Alias timemachine
HeaderDigest CRC32C,None
DataDigest CRC32C,None

Finding out what to put in the Target name (in my case, "iqn.2007-12.tld.domain.subdomain:timemachine") took me the longest, but I can recommend the documentation.
Thumbs up to Studio Network Solutions!

#11 Mr.Humbug

Mr.Humbug

    Newbie

  • Members
  • Pip
  • 5 posts

Posted 05 March 2008 - 01:12 PM

I'm using OpenFiler 2.2 and globalSAN iSCSI Initiator for OS X on a 24" iMac (model: iMac6,1) and OS X 10.5 Leopard. Sad thing is: the iMac always hangs during transfer (Time Machine). Since this is a reproducible behaviour, there is hope that further versions of the driver will finally do the trick. I've found another thread here, regarding combinations of recent Macs that work or don't work. Unfortunately, mine seems to be one of the latter. :(

#12 SNSjason

SNSjason

    Advanced Member

  • KB Moderators
  • PipPipPip
  • 36 posts

Posted 06 March 2008 - 10:40 AM

A Release Candidate for the Leopard-compatible version of globalSAN can be found here. Enjoy!

#13 Mr.Humbug

Mr.Humbug

    Newbie

  • Members
  • Pip
  • 5 posts

Posted 08 March 2008 - 09:07 AM

Too bad, the RC-1 for Leopard didn't solve the problem. The machine still hangs during transfer ... :(

I've been monitoring the last time my system started to hang and there was no unusual behviour of processes in the system monitor. There is nothing reported in the system log, except for the following message, that appears after every ~10GB chunk of data that is being transferred. It's coming from Time Machine:

Mar 5 18:30:59 localhost com.apple.backupd[321]: SystemFlippers: didn't consume all data for vers ID 1 (pBase = 0x1d377fc, p = 0x1d3780a, pEnd = 0x1d37824)

Oh, and when I toggle the "ignore owner" flag on the iSCSI volume, there is this message appearing in the system log:

Mar 6 19:17:25 localhost KernelEventAgent[46]: tid 00000000 received unknown event (256)

Also, I'm not sure if there is something wrong with the driver and the iMac's power save options. But everytime I change those settings (from "never" to "5 min." to "never"), the iMac locks up at an ammount of 2 -5 GB of data (out of a whole 145 GB) being transferred. If I don't touch the power save panel, I can go up to ~50 GB (once again, out of a whole 145 GB) of data being transferred without any problem. Maybe this issue is Time Machine related, but I don't think so.

On the iMac when the trouble starts, I can hear the harddisk perform a lot of operations. It sounds like it is indexing files or doing a file system wide search. Then, after a few seconds, the whole system freezes and can only be restarted by pressing the power switch for more than 3 seconds.

Note that the OpenFiler system is doing fine, it still keeps running and doesn't report anything in its syslog when the iMac hangs.

I hope my input helps to localize the problem and fix it once and for all - and I'm really looking forward to that moment ... ;)

P.S.: Maybe one solution would be to make some of the driver's options parametrizable, since I'm a "home user" with a "slow" 100 MBit LAN. But I feel like those hang-ups result from too large buffer sizes or I/O buffer settings and it looks like it's Intel-Mac related. But I'm definitely not an expert on this topic.

#14 pingaiter

pingaiter

    Newbie

  • Members
  • Pip
  • 8 posts

Posted 09 March 2008 - 07:23 PM

Too bad, the RC-1 for Leopard didn't solve the problem. The machine still hangs during transfer ... :(

Since installing RC-1 on two Leopard systems, it's been nothing but stable. I've tried a number of different environments: MacBook Pro wired with GigE, MacBook Pro wireless on 11g, iMac wired with GigE, and iMac wireless on 11g. The MacBook Pro has been usable even with the prior version, but GlobalSAN iSCSI initiator did not work well when the MBP went to sleep while still connected to the iSCSI target. The iMac always locked up after transferring a few GBs of data, never consistently.

RC-1 has been an entirely different story. Both macs have been pushing GBs of data over the past few days to Openfiler iSCSI targets without any problems. Over GigE I'm seeing 13MB/s transfer rates, which I'm very happy with considering it's running as a VM in VMWare Server running on a Windows 2k3 server box I had lying around.

Not sure if this is related, but the iMac had a terrible time of locking up whenever the ethernet was connected a few days after experimenting with the 3.0 iSCSI initiator... even when no target was mounted! At first I went into single user mode and ran some diagnostics, thinking there was something wrong with the system, even did an archive and install. The only thing that cured it was unplugging the ethernet cable, uninstalling GlobalSAN iSCSI initiator, uninstalling VMWare Fusion. Then I installed RC-1, and tested the target over 802.11g. Once I was confident that was working I reinstalled VMWare, and tested it out by transferring a few multi-gigabyte files back and forth. Finally I reconnected the ethernet cable which caused a hard lockup, but as soon as the system rebooted, it's been stable since. Call it dumb luck, still don't really know what went on, and I may not since I don't feel like trying to replicate the situation.

My advice is to try it over wired as well as wireless to see if the issues persist in both environments. Uninstall any other 3rd party software that also loads network related kexts (i.e. VMWare Fusion or Parallels) and test. Then reinstall and retest. See what you come up with and post here so the good folks at SNS can continue to support this much needed piece of software.

Cheers!

#15 Mr.Humbug

Mr.Humbug

    Newbie

  • Members
  • Pip
  • 5 posts

Posted 10 March 2008 - 08:09 AM

Well,

I'll give it a try and test how the globalSAN iSCSI initiator for OS X Leopard performs over a WLAN connection, although that is exactly the one thing I don't want to do: backing up via WLAN (slow connection, not really secure, don't like WLAN anyway). If that works, I'll get back to this thread. But it may take some time, since I have to once again push ~150GB of data over a 54MBit connection, just to see what happens.

As for VMWare, Parallels and other software like that: I don't use it, so I can be quite sure that there is no other kernel extension causing trouble on my system. I's just a stock Intel iMac with only a few apps (like OpenOffice, iPhoto and the other iApps).

#16 Eric Newbauer

Eric Newbauer

    Advanced Member

  • KB Moderators
  • PipPipPip
  • 254 posts

Posted 10 March 2008 - 08:32 PM

Too bad, the RC-1 for Leopard didn't solve the problem. The machine still hangs during transfer ... :(

I've been monitoring the last time my system started to hang and there was no unusual behviour of processes in the system monitor. There is nothing reported in the system log, except for the following message, that appears after every ~10GB chunk of data that is being transferred. It's coming from Time Machine:

Mar 5 18:30:59 localhost com.apple.backupd[321]: SystemFlippers: didn't consume all data for vers ID 1 (pBase = 0x1d377fc, p = 0x1d3780a, pEnd = 0x1d37824)


I believe this is a known, documented issue with Time Machine.

From the Time Machine help:
"NOTE: If Time Machine stops backing up after about 10 GB of data, check to see if your backup disk uses a Master Boot Record (MBR) partition type. If it does, you must change the partition type and erase the disk so it is supported by Time Machine. Refer to this article for more information:"

-----> http://docs.info.app...l?artnum=306932

Try the GUID partition scheme and see if that corrects the problem.
Eric Newbauer, SNS Moderator
----
Learn about our SAN/NAS shared media storage products
Follow Studio Network Solutions on Twitter

#17 Mr.Humbug

Mr.Humbug

    Newbie

  • Members
  • Pip
  • 5 posts

Posted 11 March 2008 - 12:06 PM

Try the GUID partition scheme and see if that corrects the problem.


Duh! I had set the partition type to "Apple Partition Table" since I had formatted a USB disk drive for our G4 mini recently on this system. Right now the Time Machine backup is running with only 200mb transferred. But on the openfiler box I can see a lot less activity on the internal harddisks than before while the transfer is ongoing. Before, with the wrong partition type set on the target, the disks were always busy during the whole transfer. Now it's chunk by chunk as TM sends them, just as I'm used to on a USB drive. Looks promising, but no champagne yet! I'll keep you people up to date. Thanks for the hint on the partition scheme - boy, that would have been the least thing coming to my mind ... :blink:

EDIT: well, no change at all: the 3.0 version still hangs during transfer and the 3.2 RC1 version slows down traffic to 1.5 MBit (on a 100MBit connection). So no success here, but I'll keep testing future versions of the globalSAN driver and post my results in this thread.

#18 Zeitkind

Zeitkind

    Newbie

  • Members
  • Pip
  • 6 posts

Posted 24 March 2008 - 04:43 PM

and the 3.2 RC1 version slows down traffic to 1.5 MBit (on a 100MBit connection).


Simular problem: around 3MB/s over a Gigabit connection. I'm using CentOS 5.1 with iscsitarget 0.4.16. The target is /dev/sda1 directly, an external SCSI-RAID5 using blockio. The drive (1,5TB) formatted fine as HFS+ journaled and I have no problems connecting/disconnecting/reconnecting.

ietd.conf:
Target iqn.2008-03.local.debbie:storage.sda1.raid
Lun 0 Path=/dev/sda1,Type=blockio
MaxConnections 1
InitialR2T Yes
ImmediateData No
MaxRecvDataSegmentLength 8192
MaxXmitDataSegmentLength 8192
MaxBurstLength 262144
FirstBurstLength 65536
DefaultTime2Wait 2
DefaultTime2Retain 20
MaxOutstandingR2T 8
DataPDUInOrder Yes
DataSequenceInOrder Yes
#ErrorRecoveryLevel 0
HeaderDigest CRC32C,None
DataDigest CRC32C,None

I already transfered around 3GB with Synk Pro to test stability, no problems so far except the sloooooow transfer. I can simultanously transfer files to a smb-share on the same machine and get around 20MB/s (which is much below the limit of the RAID, but it's not the fastest one..). I tested as switches a Netgear and a 3Com, no change. Both machines have cpu-load <10%. The target is a 1,6GHz Dell, the Mac is a G5 running 10.5.2.
I don't know where to look at, I tried disabling CRC, tried - just in case we have the same problem as slow samba etc - a sysctl -w net.inet.tcp.delayed_ack=0, checked interfaces for errors, but not even a single lost/resent packet.

Any hints welcome...

Gonna install SmallTreeISCSI from Atto now for a cross-check, a 30 days eval can be requested from them. Apple pulled iSCSI from Leopard (Thanx iPhone...) and I don't really like Atto, had bad experiences with their support. So I really hope SNS will keep on with their good work. :)

Edit: The Atto-Client gave exact the same transfer rate around 3MB/s. From Windows I get 35MB/s - seems to be a more basic problem on Leopard or Mac OS X or my testing machine. Strange.

#19 Zeitkind

Zeitkind

    Newbie

  • Members
  • Pip
  • 6 posts

Posted 25 March 2008 - 07:11 AM

Edit2: Small Tree Client also 3MB/s. A test from an older G4 with a gigabit-NIC and 10.4.9 gave same results. I wonder if my target CentOS + iSCSI Enterprise Linux have problems or whatever is killing the performance.
Already transfered >140GB@3MB/s with the tested 3 initiators, no problems, it's just painfully slow.

#20 SNSryan

SNSryan

    Advanced Member

  • KB Moderators
  • PipPipPip
  • 221 posts

Posted 25 March 2008 - 01:28 PM

Edit2: Small Tree Client also 3MB/s. A test from an older G4 with a gigabit-NIC and 10.4.9 gave same results. I wonder if my target CentOS + iSCSI Enterprise Linux have problems or whatever is killing the performance.
Already transfered >140GB@3MB/s with the tested 3 initiators, no problems, it's just painfully slow.

This is great feedback, thank you very much. Without having this target it's difficult to make such discoveries. Let us know if you find anything on the target side. Keep an eye on the product announcements forum ;)

regards,
ryan
EVO is an all-in-one, turnkey, SAN and NAS shared storage server specifically developed and tuned for the needs of AVID, Final Cut Pro, Adobe Premiere, Pro Tools, Autodesk Smoke, Assimilate Scratch and other film, video, graphics, VFX, animation, audio, and broadcast production software. EVO features 1/10Gb ethernet and or 4/8Gb fibrechannel - no switches needed! More info on EVO can be found here: http://www.studionet...vo-features.php