What steps will reproduce the problem?
Don't really know. Using Lusca on an ISP with 3000 users and getting this 2-3
times a week.
What is the expected output? What do you see instead?
I see this:
2011/01/10 14:15:03| storeCossCreateMemBuf: Maximum number of full buffers
reached on /usr/local/hyper/var/cache/coss1/stripe. You may need to increase
the maxfullbuffers option for this cache_dir
2011/01/10 14:15:03| assertion failed: coss/store_io_coss.c:315: "-1 !=
sio->swap_filen"
And lusca crashes.
Tried squid-2.7-stable9 with the same results...it's not lusca-only.
What version of the product are you using? On what operating system?
Linux 2.6.34, LUSCA_HEAD-r14756
Please provide any additional information below.
My coss setup is this:
cache_dir coss /coss1 65520 max-size=1048575 max-stripe-waste=32768
block-size=4096 maxfullbufs=20
cache_dir coss /coss2 65520 max-size=1048575 max-stripe-waste=32768
block-size=4096 maxfullbufs=20
cache_dir coss /coss3 65520 max-size=1048575 max-stripe-waste=32768
block-size=4096 maxfullbufs=20
mgr:info shows this:
Internal Data Structures:
6922169 StoreEntries
883 StoreEntries with MemObjects
49 Hot Object Cache Items
6921398 on-disk objects
After rebuilding, mgr:coss shows:
OPS SUCCESS FAIL
open 112106 112106 0
create 54975 54975 0
close 166997 166997 0
unlink 6720805 6720805 0
read 255877 104010 0
write 222902 222902 0
s_write 829 829 0
stripes: 33
dead_stripes: 0
alloc.alloc: 54975
alloc.realloc: 8494
alloc.memalloc: 102149
alloc.collisions: 0
disk_overflows: 0
stripe_overflows: 833
open_mem_hits: 8096
open_mem_misses: 104010
Original issue reported on code.google.com by robertpi...@gmail.com on 10 Jan 2011 at 5:20
Original issue reported on code.google.com by
robertpi...@gmail.com
on 10 Jan 2011 at 5:20