Test Environment
-
mount configuration (client)
/etc/fstab: nfs4 defaults 0 0
/proc/mounts: nfs4 rw,vers=4,rsize=65536,wsize=65536,hard,intr,proto=tcp,timeo=600,retrans=3,addr=nfs1_dmz 0 0
-
export configuration (server)
/etc/exports: *(rw,fsid=0,insecure,no_subtree_check,async)
-
network link speed
ping -s 65507 nfs-wan
65515 bytes from nfs1_dmz.frec.bull.fr (129.184.9.138): icmp_seq=17 ttl=42 time=1076 ms
65515 bytes from nfs1_dmz.frec.bull.fr (129.184.9.138): icmp_seq=19 ttl=42 time=1104 ms
65515 bytes from nfs1_dmz.frec.bull.fr (129.184.9.138): icmp_seq=20 ttl=42 time=1098 ms
65515 bytes from nfs1_dmz.frec.bull.fr (129.184.9.138): icmp_seq=21 ttl=42 time=1663 ms
65515 bytes from nfs1_dmz.frec.bull.fr (129.184.9.138): icmp_seq=22 ttl=42 time=1081 ms
65515 bytes from nfs1_dmz.frec.bull.fr (129.184.9.138): icmp_seq=23 ttl=42 time=1179 ms
65515 bytes from nfs1_dmz.frec.bull.fr (129.184.9.138): icmp_seq=24 ttl=42 time=1146 ms
65515 bytes from nfs1_dmz.frec.bull.fr (129.184.9.138): icmp_seq=25 ttl=42 time=1164 ms
65515 bytes from nfs1_dmz.frec.bull.fr (129.184.9.138): icmp_seq=31 ttl=42 time=1065 ms
The results are HERE
Nfs_fsstress tests
-
nfs_fsstress run successfully during 2hours
-
nfs_fsstress run successfully with the following parameters :
time for p in 2 4 8 16 ;
do for n in 1 2 4 8 16 32 64;
do ./fsstress -d /mnt/wan/fsstress -p $p -n $n ;
done; done;
It took:
real 41m9.808s
user 0m0.112s
sys 0m0.904s
Fsx tests
-
fsx run successfully during 2hours
-
fsx run successfully with 1000,6000 iterations and gives the following results:
time ./fsx-linux -N 1000 /mnt/bull/testfile
truncating to largest ever: 0x13e76
truncating to largest ever: 0x2e52c
truncating to largest ever: 0x3c2c2
truncating to largest ever: 0x3f15f
All operations completed A-OK!
It took:
real 17m3.548s
user 0m0.344s
sys 0m0.348s
time ./fsx-linux -N 6000 /mnt/wan/fsxfile
truncating to largest ever: 0x13e76
truncating to largest ever: 0x2e52c
truncating to largest ever: 0x3c2c2
truncating to largest ever: 0x3f15f
truncating to largest ever: 0x3fcb9
truncating to largest ever: 0x3fe96
truncating to largest ever: 0x3ff9d
All operations completed A-OK!
It took:
real 113m48.963s
user 0m0.388s
sys 0m0.916s
FFSB tests
-
ffsb run successfully during 2hours
-
ffsb run successfully with 1200 files in profile_stress
time ../ffsb profile_stress_test_wan (1200 files)
Results:
Benchmark took 307.97 sec
Total Results
===============
read : 278 ops (55.823293%)
write : 117 ops (23.493976%)
append : 103 ops (20.682731%)
1.62 Transactions per Second
It took:
real 63m31.325s
user 0m0.032s
sys 0m0.888s
Connectathon tests
-
Connectathon run successfully during 2hours
-
Connectathon run successfully with -a -f options
time ./runtests -a -f /mnt/wan/cthon04
All tests completed
It took:
real 76m25.098s
user 0m2.912s
sys 0m1.680s
-
Connectathon run successfully with -a -t options
time ./runtests -a -f /mnt/wan/cthon04
Iozone tests
-
iozone failed during 2hours test
time ./iozone -ace -r 64 -i 0 -i 1 -f /mnt/wan/iozonefile -U /mnt/wan
Here is the error we got:
VFS: Busy inodes after unmount of 0:11. Self-destruct in 5 seconds. Have a nice day...
------------[ cut here ]------------
kernel BUG at fs/nfs/nfs4state.c:126!
Issue is registered in bugzilla base at #113
Dbench tests
-
dbench run successfully during 2hours test
-
dbench run successfully for 10, 100 clients
time dbench 100
Throughput 0.143342 MB/sec 100 procs
It took:
real 69m37.743s
user 0m0.424s
sys 0m3.160s
Locktests
-
Locktests run successfully in local mode with 500 processes
./locktests -n 500 -f /mnt/wan/test_locks
process number : 500 - Remote clients: 0 local client 1 - Total client 1 - Total concurent tests: 500
process number running test successfully :
500 process of 500 successfully ran test : WRITE ON A READ LOCK
500 process of 500 successfully ran test : WRITE ON A WRITE LOCK
500 process of 500 successfully ran test : READ ON A READ LOCK
500 process of 500 successfully ran test : READ ON A WRITE LOCK
500 process of 500 successfully ran test : SET A READ LOCK ON A READ LOCK
500 process of 500 successfully ran test : SET A WRITE LOCK ON A WRITE LOCK
500 process of 500 successfully ran test : SET A WRITE LOCK ON A READ LOCK
500 process of 500 successfully ran test : SET A READ LOCK ON A WRITE LOCK
500 process of 500 successfully ran test : READ LOCK THE WHOLE FILE BYTE BY BYTE
500 process of 500 successfully ran test : WRITE LOCK THE WHOLE FILE BYTE BY BYTE
Global (nfs_stress+fsx+cthon04+ffsb+iozone) tests
-
nfs_stress + fsx + cthon04(-a-t) + ffsb + iozone were still running together simultaneously successfully when i stopped them after two hours
Software configuration
- linux-2.6.17-rc2-CITI_NFS4_ALL-1
Hardware configuration
(Server)
- Dual PIII 800MHz, cache 256KB
- Total memory: 512MB
- Ethernet: 100Mb/s link
- Distribution : gentoo
(Client)
- 2 processors : Intel(R)
- Xeon(TM) CPU 2.80GHz, cache 512 KB
- Total memory: 2Gb
- Ethernet: 100Mb/s link
- Distribution : Fedora Core
