[TriLUG] ulimit problem when trying to su - oracle
bak
bak at picklefactory.org
Fri Mar 4 20:25:52 EST 2011
A pity oracle can't be bothered to tell you which resource it can't
obtain. :)
Perhaps there is a hard kernel limit you are reaching. Even 'unlimited'
values do have some kind of a limit.
Try investigating kernel limit values using sysctl -a.
Try looking into /proc/#/fd to see how many files oracle processes have
open.
Try looking into /proc/sys/kernel (a lot of duplicates here also visible
in sysctl -a).
Try looking at /var/log/kern.log.
In the past, on Solaris, I remember having to change shm (shared memory)
and sem (semaphore) values from their defaults, or risk Oracle doing
this kind of thing.
Just some thoughts.
--bak
On 3/2/11 10:56 AM, Blackburn, Marvin wrote:
> After a new system has been up for four or five days we get the
> following error when trying to
>
> login as oracle, or su - oracle (but no other user).
>
>
>
> Resource temporarily unavailable
>
>
>
> We think this is a problem with ulimit. When we shutdown the oracle
> processes we get:
>
>
>
> # su - oracle
>
> oracle at finappsvr1> ulimit -a
>
> address space limit (kbytes) (-M) unlimited
>
> core file size (blocks) (-c) 0
>
> cpu time (seconds) (-t) unlimited
>
> data size (kbytes) (-d) unlimited
>
> file size (blocks) (-f) unlimited
>
> locks (-L) unlimited
>
> locked address space (kbytes) (-l) 32
>
> nice (-e) 0
>
> nofile (-n) 65536
>
> nproc (-u) 16384
>
> pipe buffer size (bytes) (-p) 4096
>
> resident set size (kbytes) (-m) unlimited
>
> rtprio (-r) 0
>
> socket buffer size (bytes) (-b) 4096
>
> stack size (kbytes) (-s) 10240
>
> threads (-T) not supported
>
> process size (kbytes) (-v) unlimited
>
>
>
>
>
> and doing the following commands shows
>
>
>
> oracle at finappsvr1> ulimit -Hn
>
> 65536
>
> oracle at finappsvr1> ulimit -Sn
>
> 65536
>
>
>
> oracle at finappsvr1> ulimit -Hu
>
> 16384
>
> oracle at finappsvr1> ulimit -Su
>
> 16384
>
>
>
>
>
> when I did ans lsof -u oracle during the problem, it showed just a tad
> over 15000
>
> when I looked at the process table, I only got about 100 oracle
> processes running.
>
>
>
> Is there anyway to determine what setting is causing this issue?
>
> Is there anyway to determine how much of a resource is being consumed as
> the problem is occurring.
>
> HPUX used to have a tool "GLANCE" which was sometimes helpful in this
> scenario.
>
>
>
> Any help would be helpful
>
> _____________________________________
> "He's no failure. He's not dead yet."
> William Lloyd George
>
>
>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 259 bytes
Desc: OpenPGP digital signature
URL: <http://www.trilug.org/pipermail/trilug/attachments/20110304/d2b98ac2/attachment.pgp>
More information about the TriLUG
mailing list