Subject: Re: CFS goes unresponsive under NetBSD
To: None <cfs-users@research.att.com, tech-kern@netbsd.org>
From: lcs Mixmaster Remailer <mix@anon.lcs.mit.edu>
List: tech-kern
Date: 08/03/1999 04:20:13
Copy to tech-kern because might be known kernel problem.
More detail on cfsd hang:
Last lines of ktrace/kdump output:
189 cfsd RET sendto 96/0x60
189 cfsd CALL select(0x4,0xefbfdd18,0,0,0)
Last syscall is to select (apparent via svc_run), which never returns.
System did not hang completely since one second later syslogd logs that it
has stopped responding.
Slowly programs which access near the /crypt stop working.
Programs which attempt to kill them stop working.
Had to reboot system with switch.
tcpdump on loopback during similar meltdown shows nothing special:
21:14:27.853816 localhost.xxxxxxxx21 > localhost.nfs: 128 read fh Unknown/1 8192 bytes @ 1163264
21:14:27.878854 localhost.nfs > localhost.xxxxxxxx21: reply ok 8292 read
21:14:27.879596 localhost.xxxxxxxx22 > localhost.nfs: 116 getattr fh Unknown/1
21:14:27.880399 localhost.nfs > localhost.xxxxxxxx22: reply ok 96 getattr REG 100600 ids yyy/yyy sz 10539008
21:14:27.880697 localhost.xxxxxxxx23 > localhost.nfs: 128 read fh Unknown/1 8192 bytes @ 10223616
21:14:27.881927 localhost.xxxxxxxx24 > localhost.nfs: 4228 write fh Unknown/1 4096 (4096) bytes @ 6414336 (6414336)
Transations x23 and x24 retry several times but never get a response.