expireover out of locks??

Arrgh.  On top of it mutex hanging pretty much every day now about 2am, 
expireover has been kicking out these for a few days:

May 28 01:42:52 solaris expireover[26540]: [ID 122975 news.error] OVDB: 
Lock table is out of available locks
May 28 01:42:57 solaris expireover[26540]: [ID 508361 news.error] OVDB: 
expiregroup: delete_all_records: Not enough space

I get 5-6 of them a night.  I've raised my maxlocks a couple of times from 
the default 4000.  It was 8000 tonight and happened again, and I just 
raised it to 16000 for tomorrow, but I'm suspecting that that error message 
might not be really complaining about locks.  The filesystem is only about 
50% full.

Although, maybe there are 8000 locks... I did a db_stat -CA and saw a large 
bunch of read/write locks held by the same id on various pages of ov000000, 
and pages of read locks on various handles... Maybe there were 8000 of 
them, they went on for a while.  Unfortuantely when I tried to dump them 
into a file to count them db_stat went into a mutex hang the second time I 
ran it!  Arrrgh!

My only non default ovdb settings are:

cachesize       32000
maxlocks        16000  (now; was 8000)
nocompact       0
readserver      true
numrsprocs      5

I change nocompact a week or so ago when the partition filled up, although 
that turned out to be because monitor had died and wasn't deleting log 
files.  Unfortunately I'm not sure if the problems started at the same 
time; the messages are in the "unknown entries" part of the log and are 
lost in the noise of all the spurious errors controlchan insists on 
reporting.  I haven't gotten a full expireover run in days or I'd have 
changed it back.

Interestingly, ovdb_server and monitor aren't in a mutex hang right now; 
they're either in sigsuspend or polling.