Well I imaged a computer three times in a row, and each time saving to a samba server share. When i attempt to restore the image, the md5 checksum fails, on the first part of the image (the first .AA file) I went to the samba server, and installed md5summer, and manually checked the checksums in a windows environment, and sure enough, they all check out, except the first .AA file. This was consistent across all three created images. Something I found odd, I had to change the IRP Stack size on the server before it could even successfully save a full image without reporting "insufficient space" I set the value to 40. I tried saving and image directly to a locally attached USB drive, and in this case, the image seems to have saved correctly. Something is amiss with saving to samba shares. Could this IRPStack change have somehow compromised network transfer integrity?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Maybe the problem is on network?
Is that possible you can isolate the problem by using cross-over cable to link your samba server and the machine you want to save?
Steven.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Anonymous
-
2009-11-02
Unfortunately, I can't change the wiring of the network. I did however run a full RAM test on both the client machine, and the server. There were no problems. Sucks… bad RAM would have explained everything. Anything else it could be? Sometimes I get other random behaviors, like the kernel suddenly dropping connection, and reporting a billion errors "eth0: link down". It just wont work… in a stable manner :( Anything else I could try?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
What about a new/different network card for the troubled machine? Have you imaged any other machines in the same part of the network i.e. can you rule out intervening routers, etc.?
Alan
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Anonymous
-
2009-11-02
What I can't wrap my head around is why its ALWAYS the first part of the image (.AA) and never any other part. Consistently, ONLY the first chunk fails checksum, while all other parts… ALWAYS pass. How could this be a random network problem? Its too consistent. In any case, I am trying saving images using various combinations of machines and clonezilla versions. (The newest stable clonezilla, and one old version from about a year ago.) I'l post what I find.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Yeah, I totally agree it *should* no be! :) I was just trying to think of ways to for sure rule out any possible hardware issues. Sometimes, I assume it's not a HW issue, only to have it bite me later. What about making an image w/o the Samba server, i.e. is it the server/software or machine/image?
Alan
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Anonymous
-
2009-11-02
Oddly, Whenever I tried cloning using the old version of clonezilla, it reports that the NTFS filesystem is corrupt. Every time. I tried it on 5 different machines, and happens regardless of whether i save to samba, or local device. This doesnt make sense, since this is the version of clonezilla I have been using for years! I guess microsoft changed NTFS recently? The newer version of clonezilla does not report this problem. Likely unrelated, but curious, since it is the FIRST part of the image that is always corrupted. More importantly, I created an image using a different client machine. The saved image failed checksum again. I am trying one more time, on a machine that is older and uses a different chipset. Will report what happens there. Suspect maybe the kernel version in use on this livecd may have a broken network card driver for the newer chipset?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Anonymous
-
2009-11-02
Alright the results are in. When saving image using an older machine, the image saves OK to the samba server. I saved the image twice using two different (older) client machines. The image saved fine in both cases. Therefore, the problem should not be the network, or the server machine. The problem has to do with some problem with clonezilla, and the newer chipset. I will download instead, the newer TESTING clonezilla, and see if that has the same issue.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
If you think the problem is on the kernel, maybe give Clonezilla live 20091029-karmic a try, too. It comes with different kernel. Maybe that can help you to isolate the problem.
Steven.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Anonymous
-
2009-11-05
Well i downloaded, and booted up the ubuntu based clonezilla, and encountered a totally new problem. It refuses, every time to connect to the samba server, complaining about "cannot allocate memory" and "unexpected lookup error -12". While I did encounter this problem using the other versions of clonezilla, increasing the IRPStackSize on the server machine made it work. With the new version of clonezilla, it fails to even mount the share with the same error. Every once in a while.. I notice it DOES actually mount, regardless of the error, (by examining /etc/mtab) but an attempt to save the image will reproduce the same error again later, and the image save will fail. I was not able to successfully save a single image using the ubuntu kernel. X_X I am really starting to lose patience with this :( Clonezilla seems to be the best out there for freeware disk cloning. Purchasing real software is out of the question. I really NEED clonezilla to work.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Well I imaged a computer three times in a row, and each time saving to a samba server share. When i attempt to restore the image, the md5 checksum fails, on the first part of the image (the first .AA file) I went to the samba server, and installed md5summer, and manually checked the checksums in a windows environment, and sure enough, they all check out, except the first .AA file. This was consistent across all three created images. Something I found odd, I had to change the IRP Stack size on the server before it could even successfully save a full image without reporting "insufficient space" I set the value to 40. I tried saving and image directly to a locally attached USB drive, and in this case, the image seems to have saved correctly. Something is amiss with saving to samba shares. Could this IRPStack change have somehow compromised network transfer integrity?
Maybe the problem is on network?
Is that possible you can isolate the problem by using cross-over cable to link your samba server and the machine you want to save?
Steven.
Unfortunately, I can't change the wiring of the network. I did however run a full RAM test on both the client machine, and the server. There were no problems. Sucks… bad RAM would have explained everything. Anything else it could be? Sometimes I get other random behaviors, like the kernel suddenly dropping connection, and reporting a billion errors "eth0: link down". It just wont work… in a stable manner :( Anything else I could try?
Farthom,
What about a new/different network card for the troubled machine? Have you imaged any other machines in the same part of the network i.e. can you rule out intervening routers, etc.?
Alan
What I can't wrap my head around is why its ALWAYS the first part of the image (.AA) and never any other part. Consistently, ONLY the first chunk fails checksum, while all other parts… ALWAYS pass. How could this be a random network problem? Its too consistent. In any case, I am trying saving images using various combinations of machines and clonezilla versions. (The newest stable clonezilla, and one old version from about a year ago.) I'l post what I find.
Farthom,
Yeah, I totally agree it *should* no be! :) I was just trying to think of ways to for sure rule out any possible hardware issues. Sometimes, I assume it's not a HW issue, only to have it bite me later. What about making an image w/o the Samba server, i.e. is it the server/software or machine/image?
Alan
Oddly, Whenever I tried cloning using the old version of clonezilla, it reports that the NTFS filesystem is corrupt. Every time. I tried it on 5 different machines, and happens regardless of whether i save to samba, or local device. This doesnt make sense, since this is the version of clonezilla I have been using for years! I guess microsoft changed NTFS recently? The newer version of clonezilla does not report this problem. Likely unrelated, but curious, since it is the FIRST part of the image that is always corrupted. More importantly, I created an image using a different client machine. The saved image failed checksum again. I am trying one more time, on a machine that is older and uses a different chipset. Will report what happens there. Suspect maybe the kernel version in use on this livecd may have a broken network card driver for the newer chipset?
Alright the results are in. When saving image using an older machine, the image saves OK to the samba server. I saved the image twice using two different (older) client machines. The image saved fine in both cases. Therefore, the problem should not be the network, or the server machine. The problem has to do with some problem with clonezilla, and the newer chipset. I will download instead, the newer TESTING clonezilla, and see if that has the same issue.
If you think the problem is on the kernel, maybe give Clonezilla live 20091029-karmic a try, too. It comes with different kernel. Maybe that can help you to isolate the problem.
Steven.
Well i downloaded, and booted up the ubuntu based clonezilla, and encountered a totally new problem. It refuses, every time to connect to the samba server, complaining about "cannot allocate memory" and "unexpected lookup error -12". While I did encounter this problem using the other versions of clonezilla, increasing the IRPStackSize on the server machine made it work. With the new version of clonezilla, it fails to even mount the share with the same error. Every once in a while.. I notice it DOES actually mount, regardless of the error, (by examining /etc/mtab) but an attempt to save the image will reproduce the same error again later, and the image save will fail. I was not able to successfully save a single image using the ubuntu kernel. X_X I am really starting to lose patience with this :( Clonezilla seems to be the best out there for freeware disk cloning. Purchasing real software is out of the question. I really NEED clonezilla to work.
Is that possible you can use sshfs, NFS or local disk instead of samba?
If so, please copy the image to those media, then test it again.
By doing this, we can isolate the problem…
Steven.