I was reading the recent thread and thought I may add
Using a CFS SSI 4-node cluster built using RH7.2
w/ext3 as the filesystem.
So a CFS built on ext3.
While 'gunzip-ing' a 800mb file that should inflate to
about 4 GB, the CFS kernel stops at the 2 GB limit ,
which is expected for ext2 filesystems, not ext3. To
check, I rebooted with the non-cluster kernel and the
same operation was successful and the file of 4 GB was
Now, leaving the 4 GB file there (BTW the file is a
mysql 4.0 '.MYD' table) I rebooted with the cluster
kernel, fired up the nodes, and the kernel can read
the 4 GB file OK when mysql-ing queries on the 4 GB
table. I know the entire table contents were read fine
because I had another non-cluster machine to test the
results of the table against. I also did an update on
the table so this was both read and write.
Strange is the CFS kernel can't ftp or unzip files
past 2 GB, but when a 4GB file is plopped there it can
deal with it fine.
In /etc/fstab, /dev/hda1 (The first line) is set to
Has anyone else hit a 2 GB file size limit using the
ext3 CFS kernel ? Is there a check-off I am missing in
menuconfig when I am kernel building ?
Do you Yahoo!?
Yahoo! Tax Center - forms, calculators, tips, more
Get latest updates about Open Source Projects, Conferences and News.