One of our boxes has started to do this every few nights.

 	Wondered if anyone had any thoughts as to what could be adjusted:

 	Machine is an 2GB ASUS A8V-Deluxe running NetBSD/i386 3.0
 	from 14/6/06, with a single 74GB WD raptor (its a test
 	machine hence why its not running raidframe).

 	It runs a (single) quite large apache-tomcat-5.5.14nb1 app
 	under sun-jdk15-5.0.7 (suse_base-10.0nb1), with around 150
 	webapps, and postgresql81-8.1.4 with (for some obvious
 	reason) around 150 databases.

 	I'm currently testing the native NetBSD jdk (scsl-jdk15 -
 	many thanks to tv), but as the panic occurs every few nights
 	(quite possibly during the database backups), I won't know
 	if that helps for a while.

 	panic is:

    panic: pmap_enter: no pv entries available
     Begin traceback...
     pmap_enter(a07a5920,be0a7000,4d129000,3,1) at netbsd:pmap_enter+0x3bc
     ubc_fault(x2f67c20,be0a7000,c2f67ba0,1,0) at netbsd:ubc_fault+0x20f
     uvm_fault(a0790aa0,be0a7000,0,1,c2f67e64) at netbsd:uvm_fault+0xdf6
     trap() at netbsd:trap+0x35f
     --- trap (number 6) ---
     i486_copyout()
     ffs_read()
     VOP_READ()
     vn_read()
     dofileread()
     sys_read()
     linux_syscall_plain()

 	The more interesting kernel options set are:

options 	HZ=1024			#@ ogle
options 	SHMMAXPGS=65536		#@ Postgres/ogle
options         SEMMNI=256		#@ Postgres
options         SEMMNS=1024		#@ Postgres
options         SEMMNU=256		#@ Postgres
options         SEMMAP=256		#@ Postgres

options 	NPROC=4096
options 	MAXUPRC=1024
options 	NOFILE=512
options 	NMBCLUSTERS=8192
options 	NVNODE=131072		#  128K
options 	DFLDSIZ=536870912	#  512M
options 	MAXDSIZ=2146435072	# 2047M

options 	NKMEMPAGES_MAX=98304	# Normally 32K, 128K fails to boot

# Have kernel take upper 1.5GB of VM to avoid potential issues with 4GB

options KERNBASE_LOCORE=0xa0000000
makeoptions TEXTADDR=0xa0100000

 	full kernel config & kernel at http://www.mono.org/abs/no-pv/

-- 
 			   David Brownlee -- abs@absd.org