System hang during mkfs
Hello,
I have run into a serious issue (system hang) while trying to create a luks-based loopback disk. I have reported this issue to Debian here:
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=729600
That should have most of the technical details you need to start. I suspect the behavior I'm seeing is very similar to this:
http://www.saout.de/pipermail/dm-crypt/2012-July/002544.html
I am now running memtester on most of my VM's available memory and it is now on loop #3 with no issues identified.
I'm really surprised to run across an issue like this under a stable kernel, so any advice you could give on how to further debug this would be greatly appreciated.
tim
Quoted 10 lines · Tim
Hello,
I have run into a serious issue (system hang) while trying to create a luks-based loopback disk. I have reported this issue to Debian here:
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=729600That should have most of the technical details you need to start. I suspect the behavior I'm seeing is very similar to this:
http://www.saout.de/pipermail/dm-crypt/2012-July/002544.html
Quoted 6 lines
I am now running memtester on most of my VM's available memory and it is now on loop #3 with no issues identified.
I'm really surprised to run across an issue like this under a stable kernel, so any advice you could give on how to further debug this would be greatly appreciated.
Well, the kernel Debian stable kernel, so it should be handled really handled by the Debian distro bugzilla.
(Otherwise, please try more recent, ideally 3.12 stable kernel. Using loop devices is quite common and I do not think there are recent problems here.)
Also try use loop directly. Does mkfs works ok? What is the underlying filesystem and mount parameters (on which image resides)?
If it "locks" paste kernel trace to bugzilla (through sysrq, "echo t > /proc/sysrq-trigger" and see syslog.)
(It seems to me more like problem with loop device or combination than with dmcrypt. Anyway, if you are able to reproduce it on upstream kernel, add device mapper devel list to cc - [email protected])
Milan
Quoted 6 lines
Well, the kernel Debian stable kernel, so it should be handled really handled by the Debian distro bugzilla.
(Otherwise, please try more recent, ideally 3.12 stable kernel. Using loop devices is quite common and I do not think there are recent problems here.)
That's why I submitted it there. But every single party that is involved in kernel development or support likes to point the finger at someone else without bothering to attempt reproducing. That's been my past experience, which is why I'm trying multiple avenues for support here.
Quoted 2 lines
Also try use loop directly. Does mkfs works ok? What is the underlying filesystem and mount parameters (on which image resides)?
Just tried a plain loopback device. It is a tad slower than doing mkfs directly on the file, but nothing close to the issues with the luks loopback.
Quoted 6 lines
If it "locks" paste kernel trace to bugzilla (through sysrq, "echo t >
/proc/sysrq-trigger" and see syslog.)(It seems to me more like problem with loop device or combination than with dmcrypt. Anyway, if you are able to reproduce it on upstream kernel, add device mapper devel list to cc - [email protected])
Yeah, that's where 99% of bug reporters stop...
tim