Windows 7 Version Chain fails after 3-10GB
I have three licenses for True Image and it works very well on my two XP machines. On my windows 7 laptop I am unable to get past 5GB before it hangs indefinately. Once it got to 10GB.
I am backing up to a QNAP NAS and it is not an authentication problem as it is able to write for a good 30 minutes before it hangs. I have 10TB free so space is not an issue. Has anyone else seen this?
bT
This could be timely or sleep issue on the target disk or storage enclosure.
GH42. USB prevent sleeping.
GH42A. Set WD My Passport Sleep Timer to Never Sleep.
GH43. USB Set caching/quick removal on usb disks.
Grover, that is an interesting concept. Can you elaborate? My NAS is running RAID 6 on six disks with a very fat network pipe. I can push files to it very fast using rsync or Beyond Compare and it never chokes. BTW, the disks are WD Red NAS disks if that is of any interest.
What causes the sleep issue to which you refer? The NAS does sleep but that is only after 30 minutes of inactivity. The source disk I know isn't sleeping because I am running these backups during the day when I can observe them. And, I never have this problem running TI from my Windows XP machines to the same NAS... :-(
bT
My point was to check this as an possible cause. External disks such as the WD passport have individual sleep setting.
Recently, there was a posting of the NAS or similar settings having their own sleep setting controls.
I am not familiar with your type equipment so cannot be any more specific.
There have also been suggestions that the file size be split to no larger than 10GB per slice across a network. Whether this is applicable to you, I am not sure.
Many thanks Grover. I don't think the sleep setting is an issue but I will research the slice size. For Filezilla I limit to 5GB and it works very well.
Thank you for the custom backup examples in your FAQ. One of the things I don't like about the stock version chain backup is the short retention time. I am going to try your custom incremental which retains 4 chains after my next scheduled disk clone session.
bT
bitTwiddler,
Can you provide some further details on your network configuration? How are you connecting this laptop to the NAS? Are you using USB 3.0 to connect? Are there any hubs or switches involved?
I am curious for several reasons, one is that it sounds to me like you have saturated the network with data and it is the network itself that is choking and not the TI app. Two is that I am in the planning and parts gathering stage of building a custom NAS box so I am very interested in your problem and hopefully a resolution to your issue.
It is known fact that when transferring large amounts of data across a network that transfer will slow over time. You can witness this when downloading a large file or update over the internet for example. The transfer should not stall or hang however so that is some other issue all together. I do not think that sleep is an issue here either but possibly timeout may be involved.
Can you tell me the parameters of your NIC on the laptop. What laptop make do you have and what make of NIC is it using? You obviously have other machines on the LAN that do not exhibit this problem, can you compare the NIC settings of those machines with that of the laptop NIC and note any differences?
I have a gut feeling you may have to slow down your transfer speed between your laptop and your NAS in order to make a successful completed backup of your data.
Where to begin?
It is not axiomatic that when transferring large amounts of data across a network that transfer will slow over time. It can happen for pathalogical reasons but it is not a given nor should it be. I wouldn't be so dogmatice but networks is my day job (hp fibre channel developer, etc). And networks don't choke unless there is a problem (packet storms, ip conflicts, poorly written software, etc). The network can become a bottleneck whereby the throughput cannot exceed the network's capability. The data sender merely becomes blocked on network IO.
In my case the network is not the bottleneck. I have gigabit connectivity from my laptop to the NAS. Essentially, there is one Cisco switch between my laptop and the NAS and the logs are clear WRT collisions, persistent retries, etc. Both NICs are gigabit capable. I typically get 100MBytes/sec sustained reads/writes on the QNAP.
Having said all that, I do believe there is some pathalogical network issue involving the Windows 7 laptop and the NAS. Tonight I will run Wireshark during a backup and see if I can see what is happening. Slowing the backup down may avoid hitting the issue but I would prefer to root cause if at all possible.
bT
P.S. The NAS in question is a QNAP 669-Pro.
Well, given your background you certainly should know if your network is a problem or not, that's great! I too was going to suggest that a slow down in the transfer be avoided if at all possible and I was going to suggest Wireshark as a way of checking for issues between the laptop and the NAS which by the way is a top notch device so I am confident given your stated typical sustained transfer rate is not an issue. Neither might I add would be the switch involved. Are you running an SSD in your laptop? Reason I ask is I read a post here some days ago of a user whom was having similar issues with a laptop running an SSD.
Will be most interested in what you find, please post back.
Thank you for the custom backup examples in your FAQ.
This link may be more current. I go to great lengths to avoid any use of consolidation. Using the automatiac cleanup feature plus the option "store x number of chains" works well and is a plain delete the oldest chain each time a new chain is started, so your retention is a constant rotation. the number of incs per chain, or the number of chains to keep are user adjustable based on your space or needs.
GH12. Create Custom Incremental Backup Scheme. 6 Inc, Keep 4 chains.
LOL, I am experienced but not infallible. And, every system is different. So, I am willing to entertain all suggestions at this point.
I ran Wireshark twice and experienced the same thing both times. It seems that the SMB connection is fine until it attempts to create a new slice file (-v1, v2, etc). I believe it uses this query to determine the next file name to create (-vn+1). Then a sequence of TreeConnect calls which never connect.
143151 317.236600 cumbres 192.168.0.20 SMB2 258 Create Request File: Cumbres
143152 317.238907 192.168.0.20 cumbres SMB2 242 Create Response File: Cumbres
143153 317.239194 cumbres 192.168.0.20 SMB2 260 Find Request File: Cumbres SMB2_FIND_ID_BOTH_DIRECTORY_INFO Pattern: *;Find Request File: Cumbres SMB2_FIND_ID_BOTH_DIRECTORY_INFO Pattern: *
143154 317.240426 192.168.0.20 cumbres SMB2 667 Find Response;Find Response, Error: STATUS_NO_MORE_FILES
143155 317.240632 cumbres 192.168.0.20 SMB2 146 Close Request File: Cumbres
143156 317.241155 192.168.0.20 cumbres SMB2 182 Close Response
...
143166 317.348901 cumbres 192.168.0.20 SMB2 154 TreeConnect Request Tree: \\qnap1\IPC$
143167 317.350016 192.168.0.20 cumbres SMB2 138 TreeConnect Response
143168 317.350179 cumbres 192.168.0.20 SMB2 162 GetInfo Request FILE_INFO/SMB2_FILE_STANDARD_INFO
143169 317.350596 192.168.0.20 cumbres SMB2 131 GetInfo Response, Error: STATUS_FILE_CLOSED
143170 317.350689 cumbres 192.168.0.20 SMB2 162 GetInfo Request FILE_INFO/SMB2_FILE_STANDARD_INFO
I have seen this on x86 servers when TOE is enabled so I am going to disable TOE on my NIC and see what happens. I do use the same NIC to push _many_ bits up to my HP StorServ 7000 in the office but it doesn't use SMB protocol.
Crazy!
bT
Yes, that is the one I was looking at. Good work!
Apologies guys! The trace snippet above is from me browsing the SMB shares via a file browser. Hence the directory name below the backup target. I should know not to work late on this stuff...
I just created a new backup definition and have throttled the network to 50Mbps. Stay tuned!
bT
Grover, so far you are closest to the answer yet! I may end up buying you lunch after all ;-)
I will run more tests this weekend but the heavily throttled back up is still running and is about 75% complete. Of course, 50Mbps is intolerably slow so I'll still need to perform root cause analysis. But at this point, slow is better than not at all!
bT
I saw this entry while monitoring performance this morning and it is unsettling. Has anyone see this before?
I can resolve the domain but icmp packets are blocked as traceroute cannot complete.
I can't say I have seen that before with respect to TI but it makes me question if your firewall is set to not allow icmp traffic. Could explain some of why a slow down in transmission works.
Please elaborate. I meant in my post that ICMP packets were blocked at the host hacklab.com so it couldn't be fully tracerouted. My backup doesn't go through a firewall and my firewall doesn't block ICMP and ICMP ERM packets internally - between my laptop and the NAS. As far as I know, SMB only uses ICMP ERM packets to initiate a connection. Once that is established it is not needed. Plus, the Microsoft stack will revert to NetBIOS if ICMP is blocked.
Only meant that if ICMP packets are not being passed, then if transfer rate was high enough to fragment packets, the ICMP messages indicating this would not be returned. Thus possibly Path MTU is not being detected.
Thank you for clarifying. Hopefully, this weekend will yield more clues. I noticed that the list of fixes for 2014 Update 2 sound very similar to what I am experiencing. However, my version shows it is running build 6673 and sync agent 1877.
bT
6673 is the latest build so the sync agent should be current as well. I know that NetBIOS is supposed to pass ICMP error messages but that does not always happen. So if NetBIOS fails to pass a Source Quench (4) message indicating router or host congestion then transfer slow down or stop is likely.
I am passing along a link that may be of some benefit in your troubleshooting, please refer to notation boxes at bottom of page
http://www.acronis.com/en-us/support/documentation/ATIH2014/index.html#…
Thank you for the link. FTP is a non-starter for me. I'll switch over to using another backup solution before going FTP. And the 2GB file limit gets messy as well.
ICMP is being passed. In the case of my backups, the client and the router are quiescent (no traffic outside of my backup) so normal tcp handshaking should suffice for flow control. We're not talking UDP spray-and-pray here or packet storms on the network.
bT
Got it, just never sure were a user is at so attempting to cover what bases I am aware of. I'm sure you will get it sorted out and I for one will me very interested in the outcome as I am in the planning and parts gathering stage to construct my own custom NAS server. So any hiccups I can learn about and avoid or at least deal with is a leap ahead if you know what I mean!
(offtopic) Please keep us posted on your NAS project. It sounds interesting. My QNAP is running Linux and using open source RAID software. The trickier parts are the backplane, etc - unless you are handy with a soldering iron ;-) (/offtopic)
Well, the run made it to almost 70GB before it hung. I'm going to call it a day for now...
bT
It looks like I am going to have to give up on TI for this machine. When set at a network limit of 50Mbps it runs about every other evening. The worst thing is that when it hangs it does so indefinately. Clicking the "Cancel" button does absolutely nothing. I have to reboot the machine in order to clear the hung backup for the next evening. I am surprised there is not more of an outcry on the indefinate hanging issue.
bT
This is just a notification of this posting which may or may not solve your issue.
Keep us posted if you try it.
Since this machine is behind an air-gap firewall UAC is disabled.
I was able to finally root cause the horrible performance problem. Grover, perhaps this can be added to the TI FAQ?
After a few Wireshark traces it became apparent this was clearly a Samba issue. If you search the net there is a huge amount of information around horrible Samba performance on Windows 7. After reading a few of the articles I was able to greatly improve Samba performance to the point where my backups now complete and in fairly good times as well.
Here are a few of the changes that I made:
1) Disable Autotuning
http://www.sysprobs.com/network-is-very-slow-in-windows-vista
Netsh interface tcp show global
Netsh interface tcp set global autotuning=disabled
2) Disable Remote Differential Compression
http://www.sysprobs.com/network-slow-in-vista-part-2
Remote Differential Compression (RDC) can be removed under ‘Control panel’ -> Programs and Features -> Turn Windows featured on or off.
3) DCA enabled - Direct Cache Access (DCA) – or direct cache access allows a network controller to deliver data directly to your CPU cache
netsh int tcp set global dca=enabled
4) Disable Heuristics
netsh int tcp set heuristics disabled
5) Disable NetDMA
Netsh interface tcp set global netdma=disabled
6) Change IPv4 Checksum Offload to Disabled
In the Advanced tab for my network adapter, I changed IPv4 Checksum Offload to Disabled.
I realise a few of these suggestions are controversial (NetDMA and Checksum Offload) and am interested in your thoughts on a better set of configurations. These have enabled me to backup 500MB partitions with ease.
bT
Yes, I will link to your comments. Am glad you perservered and found a solution and then took the time to post your fix for others to review.
Grover,
You are probably correct - Timeout on the external backup drives is a real issue - especially if you use the USB-3 rather than USB-2. For huge backups, you would want the former.
I just went through two months of chasing this timeout bug on THREE separate WD MY BOOK (3TB and 4TB) version drives, and they all had the same problem of invalid ATIH backups. If you use the WD or Seagate provided utility to change the USB-3 controller time to NEVER sleep, they work fine. I also found the identical timeout problem on two new SEAGATE Backup Plus (5TB) drives. But oddly, a Seagate Backup Plus (4TB), the one with 4 LED's in front panel, seemed to work best for some reason.
The problem is that ATIH will declare a backup as "Successful" in the log, until you run a "Validity" check. I trusted ATIH for a couple of years and thus never thought I needed to Validate - since all backups were deemed "successful". In fact, I eventually found they were corrupted.
It seems that the problem with WD & SEAGATE external backup drives is solely in their USB-3 "case interface" - the drive inside is perfectly fine. I suspect that during a backup, (on a regular basis), the USB-3 controller "re-checks to see" if the timeout has been exceeded (except when it is set to Never) - even though there is a heavy stream of data coming from ATIH - even for a steady couple of hours for my 600 GB FULL system backup.
For example, set to a USB-3 backup controller timeout of 15 minutes, the controller takes a peek even during busy transfers to its hard drive. That is enough time to lose a bit or several bytes of data being forwarded to the case's internal driver. The drive doesn't have a problem, since it accepts whatever the USB-3 sends to it - so no actual "disk error" is reported - the drive gets whatever comes along - good or bad and creates a correct CRC for each record (even garbled ones).
However, ATIH Validation pass checks the file recorded on the drive, and finds a flaw in transfer of data.
The reason why most customers for these "external USB-3 Backup Drives" have not been complaining, is that they use the WD or Seagate free Backup software - for their photos, movies and music, (NOT PC backup, which that software cannot do). Or they use "Microsoft Windows Backup" which comes with the OS. None of these has a "VERIFY" option of saved files. So, potentially, such "non-system type users" have corrupt and irretrievable files on such external drives.
ACRONIS is one of the very few backup programs that offers to verify a backup - and that is ONLY if the user requests it in Task setup. Perhaps, for safety, that should be the ATIH "Default".
External backup drive Manufacturers have ALL screwed this up. Only safe method is to set the USB-3 case controller to NEVER timeout (which is the factory default). Or, equally good - set the timeout to LONGER than it takes for your LARGEST PC drive backup, to avoid this seeing timeout bug.
Unfortunately, this keeps the drive spinning all the time. But ... some IT experts feel that statistically ... this is a good thing, for drive longevity.
Joe Z.
Thanks Joe for the heads-up. I have just added the link on how to adjust the WD My Passport disk.
GH42. USB prevent sleeping.
GH42A. Set WD My Passport Sleep Timer to Never Sleep.
GH43. USB Set caching/quick removal on usb disks.
Are you guys confusing this with another thread about external USB drives? This thread is about slow network performance to a NAS. ;-)