Mac client backups hang, error -505

[polymerase]

I just upgraded to Retrospect 9 on a Windows 8.1 server and I'm running into some problems completing initial backups for a few OS X 10.9 clients, running the latest Mac client version 11.0.0.191. The backup starts OK and copies tens of GB of files, then all visible progress in the activity monitor stops.  The status of the client backup shows that it's still copying files, but none of the progress indicators ever change, e.g. number of files, time. It's possible to stop the backup on the server, but the Retrospect client remains in a busy state where new backups don't start and attempts to access the client manually from the server result in the error: "Sorry, couldn't connect to <CLIENT_NAME>, error -505 (backup client reserved)".

 

I have to restart the client computer or uninstall and then reinstall the retrospet client software to regain access and restart a backup.  However, in short order the backup stalls in the same manner.  This is particularly problematic because one stalled client backup can prevent all clients in the same proactive backup script from getting backed up.

 

Has anyone else experienced this issue?

[polymerase]

Replying to my own post for an update.

 

I eventually noticed that backups for the problematic OS X clients were stalling on the same file each time, though different for each client.  When I went to the client machines the files in question were unreadable due to some disk error.

 

So, on the one hand, I discovered that some disks were going bad which is a good thing.  On the other hand, I don't think Retrospect should completely stall the backup because of a few unreadable files.  Eventually some other client's drive is going to develop some unreadable sectors and I'll have more stalled backup which require manual intervention to resolve.  As an aside, Time Machine backups do finish successfully.  I don't recall this behavior with our previous version of Retrospect (Win, v7).

[Scillonian]

Unless the network connection to a client is physically broken I think Retrospect will wait indefinitely if it has asked the client is do something and is waiting for it to respond on completion of the task. I've had clint communication hang on Windows clients due to flaky OEM provided network drivers on the client. (I don't have any Apple clients.)

 

Windows 7 used SMBv2 while Windows 8.1 uses SMBv3. From what I can find from OSX 10.9 Apple started to use SMBv2.

 

The Microsoft Knowledge Base article http://support.microsoft.com/kb/2696547 gives details on how to disable specific SMB versions for troubleshooting. This could help to find out if this is a Retrospect problem or a network communication problem.

[kleeberg]

Interestingly, I am having the same problem but I am using Retrospect on a Mac OS X 10.10.1 Server. I am running Retrospect 10.5.3 and client 11.5.0 and the backup hangs on one Mac running 10.10.1 on the network. As with the first poster, Retrospect hangs and the process needs to be stopped on the server but the client causing the hang must be rebooted else I get the -505 error. Assessing the disk with disk utilities does not show a problem. Time machine works just fine. This is the first time I have noted the file which caused the hang and I will see if it is consistent. I also believe this first appeared after upgrading Retrospect to 10.5.x

 

It troubles me that Retrospect would hang as a result of a faulty client and that the only way I discovered the problem was when I received notification on my laptop that it had not been backed up in 7 days. Retrospect has worked so flawlessly for me for years that it had become a "set-it-and-forget-it" app. I will address my problem with the local client, but I believe Retrospect needs to address this issue with, at a minimum, notification if it appears that a client has hung.

 

Paul

[kleeberg]

I am baffled. So far, the Retrospect console diplays three different files (one twice so far) when it freezes:

 

receipt.0.cdt

C079F490-7164-45F4-ACC4-88F9E6C4D02A.1.cdt

receipt.0.cdt

57CA01DD-E52A-4D09-9AAB-7BC3D7CEDB23.1.cdt

 

Search of the client using Spotlight and including the system files did not reveal any files with the .cdt extension. Running "locate" in terminal both before and after running "locate.updatedb" did not reveal any files with the .cdt extension. Searching the web indicates that this file belongs to CorelDraw but that app has ever been on this client. Console also does not reveal anything with the .cdt extension. The retrospect log does not reveal any information since, when I stop the activity, the log only shows "Script 'XXXX' stopped by operator"

 

Thoughts?

 

Paul

[kleeberg]

Found the location of these .cdt files by searching google for receipt.0.cdt but still unclear as to their purpose. Used terminal to manually delete the ones I could find (interestingly wildcards did not work in these directories) but retrospect still chocked on some other oddly named .cdt files i could not find. Partitioned the drive, used Carbon Copy Cloner to clone the drive onto the new partition and the backup went smoothly.

 

Still I think 1) a fault client should not freeze the backups, 2) if activity stalls, an alert should be triggered, 3) the path of the offending file should be able to be retrieved in a log someplace.

 

Paul

[karma]

I had this same problem recently. Hanging on various .cdt files that cannot be found doing a search on the computer. After some research I discovered that a .cdt file is related to iCloud. In my case, I can directly tie this to the fact that I turned off iCloud on the computer in question, and started getting these hangs while backing up the computer immediately after. I fixed this by deleting the computer's ~/Library/Mobile Documents.xxxx folder (xxxx is a random number, and this is where the .cdt files live, even though they can't be located by a search), and the ~/Library/Application Support/Ubiquity folder (following some instructions in a related google thread), and restarting the computer.