Mountain Lion support

[Interstellar]

Guys who have installed ML GM, UnRAID all good still?

[mrcsx]

No problems here!  I've been running the last few betas and the gm on my retina MacBook Pro with no issues with unraid!

[davidw4806]

I loaded Mountain Lion yesterday and now my AFP shares are dropping.  I made sure the Turn drives off option is not on in the preferences but it appears to lose the connection when the Mac goes into screen sleep mode.  The computer is set to never go to sleep itself.

 

I have sat with the computer for an hour and it does not lose the connection, only when I leave it unattended does the screen blank and the AFP dies.

 

So far that is the only problem I have had and I have not pinned it down exactly just surmising the issue.

[davidw4806]

AFP is unstable on my Mac now that I have upgraded to Mountain Lion Server.  I have 5.0-RC5 and it was bullet proof under Lion, it would stay connected for days and transfer TB's of data but now it is lucky to run until I have 70GB transferred.  Not sure what it is but something changed on the Mac.

[davidw4806]

I have switched to SMB and the copying has been bullet proof.  Since I am in the process of copying 1 TB of data I will not be able to do any more research until sometime next week.  It does appear Apple changed something in their AFP protocol or processes that is not handled properly with Nettalk.

[davidw4806]

Jul 26 23:47:17 Tower afpd[1940]: Terminated disconnected child[5684], client rebooted.

Jul 26 23:54:25 Tower afpd[1940]: Terminated disconnected child[15273], client rebooted.

Jul 26 23:56:40 Tower afpd[1940]: Reading IPC header failed (-1 of 14 bytes read): Connection reset by peer (Minor Issues)

Jul 26 23:58:29 Tower afpd[1940]: Terminated disconnected child[15613], client rebooted.

Jul 26 23:58:29 Tower afpd[1940]: Terminated disconnected child[7576], client rebooted.

Jul 26 23:59:19 Tower afpd[1940]: Reading IPC header failed (-1 of 14 bytes read): Connection reset by peer (Minor Issues)

 

I grabbed this right after my Macbook told me the Tower Server had shutdown.  Of course it had not shutdown and the SMB copy is still continuing on the other Mac.  Maybe this will help debug the problem.

[mle]

I've noticed the same thing (or a variation of it). What I've noticed was that automatic backups never connect to unRaid, I found an error message describing that when I return. If I run a backup manually, it will usually complete just fine as long as I babysit it. I haven't thoroughly tested it as you have but what you're experiencing sounds much like what I am after the upgrade.

 

Post that on Apple forums too if you haven't already. There are bunch of people with Time Capsules reporting very similar issues after upgrading to ML. Perhaps there's a slight chance it will get Apple's attention on there. It seems to clearly be a bug. I feel grateful, a lot of people with time capsules can't get backups to run at all. I can at least get them to run manually albeit unreliably.

Here's one (there's more):

https://discussions.apple.com/message/19064016?ac_cid=tw123456#19064016

[dgaschk]

Have you tried this: http://jonathansblog.co.uk/automount-network-on-startup-drive-osx-lion Enable "Connected Servers" in Finder Preferences first. Once the share icon is in Startup Items, Connected Servers can be unchecked.

[stealth82]

Jul 26 23:47:17 Tower afpd[1940]: Terminated disconnected child[5684], client rebooted.

Jul 26 23:54:25 Tower afpd[1940]: Terminated disconnected child[15273], client rebooted.

Jul 26 23:56:40 Tower afpd[1940]: Reading IPC header failed (-1 of 14 bytes read): Connection reset by peer (Minor Issues)

Jul 26 23:58:29 Tower afpd[1940]: Terminated disconnected child[15613], client rebooted.

Jul 26 23:58:29 Tower afpd[1940]: Terminated disconnected child[7576], client rebooted.

Jul 26 23:59:19 Tower afpd[1940]: Reading IPC header failed (-1 of 14 bytes read): Connection reset by peer (Minor Issues)

 

I grabbed this right after my Macbook told me the Tower Server had shutdown.  Of course it had not shutdown and the SMB copy is still continuing on the other Mac.  Maybe this will help debug the problem.

It happens to me when copying back any data for a total over 3GB. At times the copy hangs/pauses, then it continues for a while (If I'm lucky to 4.5GB):

 

Mac OS X Console log

03/08/12 14:16:18,000 kernel[0]: AFP_VFS afpfs_DoReconnect started /Volumes/tm prevTrigger 1 currTrigger 2
03/08/12 14:16:18,000 kernel[0]: AFP_VFS afpfs_DoReconnect:  doing reconnect on /Volumes/tm
03/08/12 14:16:18,000 kernel[0]: AFP_VFS afpfs_DoReconnect:  soft mounted and hidden volume so do not notify KEA for /Volumes/tm
03/08/12 14:16:18,000 kernel[0]: AFP_VFS afpfs_DoReconnect:  Max reconnect time: 30 secs, Connect timeout: 15 secs for /Volumes/tm
03/08/12 14:16:18,000 kernel[0]: AFP_VFS afpfs_DoReconnect:  connect to the server /Volumes/tm
03/08/12 14:16:18,000 kernel[0]: AFP_VFS afpfs_DoReconnect:  Logging in with uam 1 /Volumes/tm
03/08/12 14:16:18,000 kernel[0]: AFP_VFS afpfs_DoReconnect:  Restoring session /Volumes/tm

 

 

Then it disconnects without any valid explanation. Impossible to copy any data if not manually splitting the job into chunks smaller than 3/4GB. A PITA if you wanted to install OS X from scratch...

 

The funny thing is that I don't have this kind of problems when I'm backing up to Time Machine.

 

What netatalk version is unRAID RC6 currently using? Can we expect to see that package upgraded to 3.x? Maybe it could fix things...

 

Other lines I get:

 

03/08/12 15:05:14,000 kernel[0]: ASP_TCP ReplayPendingReqs: replaying slot 40 with reqID 37979 afpCmd 0x11 on so 0xffffff8024c0ccb0
03/08/12 15:05:14,000 kernel[0]: ASP_TCP ReplayPendingReqs: replaying slot 41 with reqID 37980 afpCmd 0x3C on so 0xffffff8024c0ccb0
03/08/12 15:05:14,000 kernel[0]: ASP_TCP ReplayPendingReqs: replaying slot 42 with reqID 37981 afpCmd 0x3C on so 0xffffff8024c0ccb0
03/08/12 15:05:14,000 kernel[0]: ASP_TCP ReplayPendingReqs: replaying slot 43 with reqID 37982 afpCmd 0x3C on so 0xffffff8024c0ccb0
03/08/12 15:05:14,000 kernel[0]: ASP_TCP ReplayPendingReqs: replaying slot 44 with reqID 37983 afpCmd 0x3C on so 0xffffff8024c0ccb0
03/08/12 15:05:14,000 kernel[0]: ASP_TCP ReplayPendingReqs: replaying slot 45 with reqID 37984 afpCmd 0x3C on so 0xffffff8024c0ccb0
03/08/12 15:05:14,000 kernel[0]: ASP_TCP ReplayPendingReqs: replaying slot 46 with reqID 37985 afpCmd 0x3C on so 0xffffff8024c0ccb0
03/08/12 15:05:14,000 kernel[0]: ASP_TCP ReplayPendingReqs: replaying slot 47 with reqID 37986 afpCmd 0x3C on so 0xffffff8024c0ccb0
03/08/12 15:05:14,000 kernel[0]: ASP_TCP ReplayPendingReqs: replaying slot 48 with reqID 37987 afpCmd 0x3C on so 0xffffff8024c0ccb0
03/08/12 15:05:14,000 kernel[0]: ASP_TCP ReplayPendingReqs: replaying slot 49 with reqID 37989 afpCmd 0x22 on so 0xffffff8024c0ccb0
03/08/12 15:05:14,000 kernel[0]: AFP_VFS afpfs_DoReconnect:  get the reconnect token
03/08/12 15:05:14,000 kernel[0]: ASP_TCP ReplayPendingReqs: replaying slot 71 with reqID 347 afpCmd 0x11 on so 0xffffff8024c00de0
03/08/12 15:05:14,000 kernel[0]: AFP_VFS afpfs_DoReconnect:  get the reconnect token

 

I still am not sure this is related with AFP but I'm also noticing sever smartctl defunct processes concurrently the hanging of the copy from Time Machine.

[stealth82]

Additional information:

 

when the copy hangs sending the ls command from terminal inside the tm folder (e.g. /mnt/disk1/tm/iMac.sparsebundle/bands) makes the terminal hang as well. When eventually Time Machine gives up copying a big file the terminal with its launched ls command comes back to life and I can actually see the folder content.

 

ATM I'm screwed very badly. After a couple of years I reinitialized OS X from scratch, the backup from Time Machine is there and I cannot have data over 5GB back. Unfortunately files over that size can't be split, so my iPhoto Library and VMs are actually lost.

 

I think this is a problem that must be fixed. I know it's not an unRAID problem but if the feature doesn't work consistently it shouldn't even be possible to enable it.

 

A serious bug fixed with netatalk 2.1.3: https://bugs.launchpad.net/ubuntu/+source/netatalk/+bug/698107

https://bugs.launchpad.net/ubuntu/+source/netatalk/+bug/698107

But it seems that unRAID already sports 2.2.3. Am I wrong?

[stealth82]

All right, this is my last post with the weirdest workaround ever:

 

Since I started noticing either afpd or OS X at a certain point dropped the connection while copying (the copy is clearly waiting for something), as I already wrote, I also found that a simple ls command under the following path - /mnt/disk1/tm/iMac.sparsebundle/bands - hung as well. I wanted to see during the copy for how long I could issue the ls command before the afpd dropped the connection.

Well, guess what! It didn't drop it as long as I kept issuing the ls command (at 1 second or less of interval). I was able to copy my iPhoto library (11GB) and my VMs (35GB) and it worked! 15 minutes of ls over and over again.

 

Now, please, somebody explains me that...

[dgaschk]

Try disabling spin-down. It may be a driver or HBA firmware issue.

[stealth82]

How can that be possible?

I'm using TM on a disk that virtually never spins down (Plex's fault here) and the problem occurs after 40/50 seconds the copy is started. Moreover I should see in the syslog attempts from unRAID to spin the disk down, am I wrong?

[dgaschk]

The system issues spindown requests. If the disk is busy it should be ignored. It can cause issues depending on the system specs, firmware revisions, and unRAID beta version.

[stealth82]

I don't think that is the problem. I can 'ls' any other folders besides the iMac.sparsebundle/bands and no hangs occur in the meanwhile.

[dgaschk]

What happens if you "ls" on a different disk while a transfer is occurring.

[stealth82]

I can ls on a different folder of the very same disk TM is getting data from with no problems at all

[dgaschk]

I can ls on a different folder of the very same disk TM is getting data from with no problems at all

 

What are the complete hardware specs and nRAID version?

[stealth82]

unRAID Version: unRAID Server Plus, Version 5.0-rc6-r8168-test

Motherboard: ZOTAC - H67ITX-C-E

Processor: Intel® CoreTM i3-2100T CPU @ 2.50GHz - 2.5 GHz

Cache: L1 = 128 kB  L2 = 512 kB  L3 = 3072 kB 

Memory: 4 GB - DIMM0 = 1333 MHz  DIMM4 = 1333 MHz  DIMM7 = 342 MHz 

Network: 1000Mb/s - Full Duplex

[dgaschk]

unRAID Version: unRAID Server Plus, Version 5.0-rc6-r8168-test

Motherboard: ZOTAC - H67ITX-C-E

Processor: Intel® CoreTM i3-2100T CPU @ 2.50GHz - 2.5 GHz

Cache: L1 = 128 kB  L2 = 512 kB  L3 = 3072 kB 

Memory: 4 GB - DIMM0 = 1333 MHz  DIMM4 = 1333 MHz  DIMM7 = 342 MHz 

Network: 1000Mb/s - Full Duplex

 

How many drives? Any SATA cards?

[stealth82]

No additional SATA controllers. My mainboard has 6 built-in SATA ports. I'm using 6 disks.

 

4 WDC_WD20EARS

2 SAMSUNG_HD204UI

[p1lot]

I only plan to use AFP for streaming Movies because SMB seems to be unreliable on my ned MacBook Pro. Do the above problems only occur when copying files to the server or also when streaming? I am running RC5.

[IsSidha]

i have been using AFP for quite some time without any serious problems..  i have iTunes streaming to my ATV's and i can copy over files to the Tower without hickups..

the only trouble i came across was my automatic startup script.. it would call the remote share and mount it, but that doesn't work.. it gives an No. -1069 error and quits.. using the method of dragging the drive into the login items works..

ofcourse, i am not copying over terrabytes of data so that may be a difference

 

i had a testrun going with freenas not so long ago, and it had the same problems as i recall... after a short bit the transfer would hang and not go anywhere anymore..  i was using a developer edition of mountain lion at the time..

[EdgarWallace]

@stealth82, how did you fixed your issue? Or did you gave up? I'm having the same issues running Mountain Lion:

 

Tower afpd[7809]: Terminated disconnected child[20863], client rebooted.

 

Since Sep. 19th Time Machine isn't running any more  I'm pretty sure that this was the day I updated from 10.8.1 to 10.8.2...

 

Anyone else with the same issue?

[stealth82]

No solution so far. I think it's OS X release independent. The problem very likely is on the afpd's shoulders.