Hi Trond-
I’ve found that after 4.1-rc1, cthon basic test 5 fails when run over
NFSv4.0 against Solaris servers (affects both TCP and RDMA).
> [cel@manet cthon04]$ ./server -b
>
> sh ./runtests -b -t /mnt/lebasque/manet.test
>
> Starting BASIC tests: test directory /mnt/lebasque/manet.test (arg: -t)
>
> ./test5a: write
> ./test5a: (/mnt/lebasque/manet.test) 'bigfile' has size 0, should be 1048576
> basic tests failed
> Tests failed, leaving /mnt/lebasque mounted
> [cel@manet cthon04]$
I bisected to commit 5bb89b4702e2 ("NFSv4.1/pnfs: Separate out metadata
and data consistency for pNFS”). I’ve been comparing network captures
and I don’t see anything that stands out.
Reverting just the change to nfs_wb_all() makes the regression go away.
As a rough guess, I think the problem could be that Solaris is handing
out a write delegation, and after commit 5bb89b4702e2, the client is not
flushing as often as it needs to ?
--
Chuck Lever
chuck[dot]lever[at]oracle[dot]com
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
