FUJITA Tomonori schrieb:
On Tue, 23 Dec 2008 11:49:45 +0100
Tomasz Chmielewski <mangoo@xxxxxxxx> wrote:
FUJITA Tomonori schrieb:
I guess that we create too many pthreads per lun. Probably, it would
be better to have some pthreads per target (regardless of the number
of luns). But it takes some time to fix the second problem.
Any news on this?
I'll be hitting this limit (~60 targets and more) on my 32 bit targets
by the end of this year and this issue makes me feel uncomfortable :(
I don't loosen the limit yet but this patch should fix the segfault
and deadlocks caused by the target or lun creation. Can you try this?
Hmmm...
# tgtd -f
can't open /proc/sys/fs/nr_open, No such file or directory
You don't enable procfs?
I do, but I don't have this file on my system (running 2.6.24.3):
# ls /proc/sys/fs/
aio-max-nr dentry-state file-max inode-nr inotify/
leases-enable nfs/ overflowuid
aio-nr dir-notify-enable file-nr inode-state lease-break-time
mqueue/ overflowgid suid_dumpable
I think it was added in a later kernel.
Then, can you disable nr_file_adjust() in the patch and try again?
(...)
lun 40
lun 41
lun 42 <- output from the script
tgtadm: out of memory <- tgtadm output
Existing targets still work after that, but tgtd won't accept any new
targets/luns.
--
Tomasz Chmielewski
http://wpkg.org
--
To unsubscribe from this list: send the line "unsubscribe stgt" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
