Description of problem: kernel hangs when writing multiple files simultaneously
on a filesystem which remains on a encrypted loop device (represented by a file
which remains a on a NFS share).
Version-Release number of selected component (if applicable): 2.6.9-42.0.3.ELsmp
How reproducible: always, the bug doesn't trigger if you just read files
Steps to Reproduce:
1. Use the attached script to create a large enough FS image (on our production
server it's a 200GB file)
2. Use any other script or application which is able to create and write at
least ten files _concurrently_ in different directories (we have four level nesting)
3. Under such circumstances our server becomes entirely unresponsive, it either
hangs instantly, or it runs for around hour or two without allowing writing any
files to any HDDs - so no SSH logins, crond tasks cannot start ... then it
hangs. If you are logged on at this time - you cannot even run top or ps - they
both hang at startup.
Actual results: hang
Expected results: work
1) I suppose this bug is not hardware related. This server (two Prescott based
Xeon CPUs) is located in a server room where temperature doesn't ever exceeds 20
2) Usually when this server hangs a load average value is around 70-200.
Created attachment 150068 [details]
A script in use
You tried to change the Status field from NEW to CLOSED, but only the owner or
submitter of the bug, or a autorized user, may change that field.
(In reply to comment #2)
> You tried to change the Status field from NEW to CLOSED, but only the owner or
> submitter of the bug, or a autorized user, may change that field.
Why do you want to close the bug?
Due to frequent kernel crashes, we retracted from using this solution. Besides
in seven (!) months you are the first person to notice this bug. My employer
couldn't wait for seven months - since that we implemented a different solution.
Bug was not assigned properly. kernel-maint is for the Fedora kernel.