Re: NFS issues on BETA7

* gnn_at_freebsd.org <gnn_at_freebsd.org> [20041014 14:19]:
> At Thu, 14 Oct 2004 13:07:52 +0200, Joan Picanyol wrote:
> > What else should I do to debug this further?
> 
> Can you tell us more about system load? 

About 0.

> Is this easy to reproduce?

For me is at's easy as starting Gnome and trying to start Evolution and
Nautilus, with /home NFS mounted (although they never show up). That
will send gconfd to lalaland, and make it unkillable.

> Is it intermittent?

No, once started everything starts braking. i.e.: attempting a login on the
text console gives:

load: 0.05  cmd: bash 1107 [getblk] 0.00u 0.14s 0% 1892k

or:

load: 1.28  cmd: bash 1115 [nfs] 0.01u 0.13s 0% 1888k

The traces for these two are a little bit different:

db> tr 1107
sched_switch(c1c3f960,0,1,11d,d5492d24) at sched_switch+0x170
mi_switch(1,0,c0681b32,19d,1) at mi_switch+0x2ac
sleepq_switch(cbf1c480,c067e375,185,1,da7299c0) at sleepq_switch+0x134
sleepq_wait(cbf1c480,0,c067f742,db,0) at sleepq_wait+0x41
msleep(cbf1c480,c06d3cc8,50,c068638e,0) at msleep+0x3b5
acquire(da729a1c,2000020,600,e7,c1c3f960) at acquire+0xb0
lockmgr(cbf1c480,2090022,c23ca528,c1c3f960,c06ad35c) at lockmgr+0x41a
getblk(c23ca528,1,0,2000,0) at getblk+0x160
nfs_getcacheblk(c23ca528,1,0,2000,c1c3f960) at nfs_getcacheblk+0x107
nfs_bioread(c23ca528,da729c80,a0000,c2013980,c22dc154) at nfs_bioread+0x788
nfs_read(da729c0c,1020002,c1c3f960,219,da729c80) at nfs_read+0x41
vn_read(c22dc154,da729c80,c2013980,0,c1c3f960) at vn_read+0x198
dofileread(c1c3f960,c22dc154,3,80f2000,23231) at dofileread+0xa4
read(c1c3f960,da729d14,c,28247000,3) at read+0x6b
syscall(2f,2f,2f,8098dc0,3) at syscall+0x272
Xint0x80_syscall() at Xint0x80_syscall+0x1f
--- syscall (3, FreeBSD ELF32, read), eip = 0x281c5ecf, esp = 0xb

db> tr 1115
sched_switch(c1c22c80,0,1,11d,df189f24) at sched_switch+0x170
mi_switch(1,0,c0681b32,19d,1) at mi_switch+0x2ac
sleepq_switch(c23ca5d4,c067e375,185,1,da69497c) at sleepq_switch+0x134
sleepq_wait(c23ca5d4,0,c067f742,db,0) at sleepq_wait+0x41
msleep(c23ca5d4,c06d4004,50,c068ac16,0) at msleep+0x3b5
acquire(da6949d8,1000000,600,e7,c1c22c80) at acquire+0xb0
lockmgr(c23ca5d4,1010002,c23ca528,c1c22c80,da6949f4) at lockmgr+0x41a
vop_stdlock(da694a10,da694a2c,c05525cb,da694a10,c06875bf) at vop_stdlock+0x2c
vop_defaultop(da694a10,c06875bf,da694a2c,c04da3da,c23ca528) at vop_defaultop+0x1
8
vn_lock(c23ca528,10002,c1c22c80,7dc,c06875bf) at vn_lock+0x10b
vget(c23ca528,2,c1c22c80,c06e2580,0) at vget+0x106
nfs_lookup(da694b90,c1c22c80,c1c22c80,c1c22c80,c1c22c80) at nfs_lookup+0x2a5
lookup(da694c28,c1c16800,0,a4,c1c22c80) at lookup+0x32e
namei(da694c28,c0682602,69f,c06ff3dc,da694c54) at namei+0x272
stat(c1c22c80,da694d14,c069735e,3df,2) at stat+0x52
syscall(2f,2f,2f,1,80d6500) at syscall+0x272
Xint0x80_syscall() at Xint0x80_syscall+0x1f
--- syscall (188, FreeBSD ELF32, stat), eip = 0x281c51ef, esp = 0xbfbfe90c, ebp
= 0xbfbfe988 ---


I've rebuilt with WITNESS and INVARIANTS (just in case), and will
recvsup tomorrow. Any directions towards solving this issue much
appreciated.

tks
--
pica