fix illogical behavior in balance_dirty_pages()
I think I'm seeing the same VM behavior with NFS/ext3 that was described with fuse.
I have a shared storage unit with 2 servers. The local filesystem is ext3 and I'm failing the ext3 mounting and NFS serving from one node to another. I'm running I/O on the NFS
mount point to the active server. If I failover to the server that is also executing the I/O,
I can hang the machine. Looking at the stack traces, the NFS writes are stuck in balance_dirty_pages(). NFS start up is causing rpc.mountd to write to the local ext3 filesystem, but it gets blocked in balance_dirty_pages() too. Deadlock. Without the NFS server successfully started, the kernel can't write the dirty pages.
I'm running a 2.6.9-42 RH kernel. Early in my debugging, I had tried several kernel versions. All 2.6 kernels saw this behavior and the 2.4 kernel I tried did not.
Kris Corwin