This case is the sole reason I have had to skip upgrading to Lucid. And as the problem seems to be also in Maverick, it seems that I will be stuck with Karmic for even longer.
Please find below a recapture of my findings, hope it would help in investigating this.
I can reliable repro this case both with the latest Maverick beta (2.6.35-22) and Lucid (2.6.32-24). I have only tested with amd64.
Here are the minimal steps needed in my environment
2. Export a nfs mount in host, mount that in guest. I use the following settings:
server: rw,nohide,insecure,no_subtree_check,async
client: timeo=14,rw,_netdev
3. Run the following script in guest, the script hangs every time, latest after creating a few thousand files.
cd <insert-your-nfspath-here>
while true
do
for i in $(seq 6)
do
dd bs=1k count=8000 if=/dev/zero of="file_"$k" &
k=$((k+1))
done
wait
done
This case is the sole reason I have had to skip upgrading to Lucid. And as the problem seems to be also in Maverick, it seems that I will be stuck with Karmic for even longer.
Please find below a recapture of my findings, hope it would help in investigating this.
I can reliable repro this case both with the latest Maverick beta (2.6.35-22) and Lucid (2.6.32-24). I have only tested with amd64.
Here are the minimal steps needed in my environment
1. Amd64 Maverick (or Lucid) host, install amd64 Maverick (or Lucid) kvm guest.
2. Export a nfs mount in host, mount that in guest. I use the following settings: insecure, no_subtree_ check,async
server: rw,nohide,
client: timeo=14,rw,_netdev
3. Run the following script in guest, the script hangs every time, latest after creating a few thousand files.
cd <insert- your-nfspath- here>
while true
do
for i in $(seq 6)
do
dd bs=1k count=8000 if=/dev/zero of="file_"$k" &
k=$((k+1))
done
wait
done