out of virtual shared memory ; buffers not flushed
Posted in 2004
Topics: Error Codes & Troubleshooting, Logging & Checkpoints
Dear
All,
I have the following log information which complain about out of virtual
memory. Obtain from onstat -l command shows the following:
23:27:23 Maximum server connections 120
23:29:34 Requested shared memory segment size rounded from 1500000KB to
1500160KB
23:29:52 Dynamically allocated new extension shared memory segment (size
1500160KB)
23:32:38 Fuzzy Checkpoint Completed: duration was 1 seconds, 7145 buffers not
flushed.
23:32:38 Checkpoint loguniq 197, logpos 0xb5b02bc
23:32:38 Maximum server connections 120
23:37:51 Fuzzy Checkpoint Completed: duration was 5 seconds, 7145 buffers not
flushed.
23:37:51 Checkpoint loguniq 197, logpos 0xb5cd2bc
23:37:52 Maximum server connections 120
23:39:31 Requested shared memory segment size rounded from 1500000KB to
1500160KB
23:39:31 shmget: [ENOMEM][12]: key 5256481d: out of shared memory, check
system max shared memory segment size
23:39:31 out of virtual shared memory
23:39:31 Requested shared memory segment size rounded from 1500000KB to
1500160KB
23:39:31 shmget: [ENOMEM][12]: key 5256481d: out of shared memory, check
system max shared memory segment size
23:39:42 Requested shared memory segment size rounded from 1500000KB to
1500160KB
23:39:42 shmget: [ENOMEM][12]: key 5256481d: out of shared memory, check
system max shared memory segment size
..
09:49:07 Maximum server connections 120
09:49:12 Assert Failed: Exception Caught. Type: MT_EX_OS, Context: mem
09:49:12 IBM Informix Dynamic Server Version 9.30.FC6X6
09:49:12 Who: Session(474164, metrica@mtcmetrica, 26188, -277102952)
Thread(502903, sqlexec, 1e7add168, 9)
File: mtex.c Line: 377
09:49:12 Action: Please notify IBM Informix Technical Support.
09:49:12 stack trace for pid 6083 written to /tmp/af.b05feb97
09:49:13 See Also: /tmp/af.b05feb97, shmem.b05feb97.0
09:49:25 Assert Failed: Exception Caught. Type: MT_EX_OS, Context: mem
09:49:25 IBM Informix Dynamic Server Version 9.30.FC6X6
09:49:25 Who: Session(474156, metrica@mtcmetrica, 26180, -632051656)
Thread(502895, sqlexec, 1f2e92088, 3)
File: mtex.c Line: 377
09:49:25 Action: Please notify IBM Informix Technical Support.
09:49:25 stack trace for pid 6077 written to /tmp/af.b057eba5
09:49:26 See Also: /tmp/af.b057eba5, shmem.b057eba5.0
10:02:25 Fuzzy Checkpoint Completed: duration was 1 seconds, 7151 buffers not
flushed.
10:02:25 Checkpoint loguniq 197, logpos 0xb6fa2ec
My question is can command : onstat -F
able to resolve this problem.
Comments and help is much appreciated.
Thanks, Patrick.
----LNX_Thu_Aug_12_2004_09:59:36_V3.33--
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
>Datum: 2004.08.12 08:43:24
>Sender: PATRICK LEY <patrickley@excite.com>
>
>Dear All,
>I have the following log information which complain about out of virtual=
>memory. Obtain from onstat -l command shows the following:
>
>23:27:23 Maximum server connections 120
>23:29:34 Requested shared memory segment size rounded from 1500000KBto 1=
500160KB
>23:29:52 Dynamically allocated new extension shared memory segment (size=
1500160KB)
>23:32:38 Fuzzy Checkpoint Completed: duration was 1 seconds, 7145 buffer=
s not flushed.
>23:32:38 Checkpoint loguniq 197, logpos 0xb5b02bc
=
The 'buffers not flushed' is normal with fuzzy checkpoints. You can disab=
le fuzzy checkpoints
with ONCONFIG-parameter
NOFUZZYCKPT 1
to be super safe (SAP recommends that parameter to be set).
=
onmode -F will not help you with the shared memory segments, because it o=nly frees unused
segments. Since your server allocates more and more segments it needsthem=
=2E Therefore there
is nothing that could be freed with onmode -F.
=
I presume a session uses more an more memory. You will have to monitor th=
e size off the
sessions with
onstat -g sesIf you do that periodically (e.g. cron) you should see one or more sessio=
ns growing constantly.
You can monitor other memory usage with onstat -g mem , too.
=
We have seen the following causes for very high memory consumptions:
recursive procedure calls
usage of hold cursors in a loop without commits (?bug? in IDS)
=
Regards,
Andreas Kutsche
------------------------------------------------
SPAR Oesterreichische Warenhandels-AG
Hauptzentrale
Europastra=DFe 3
A-5015 Salzburg
=
Telefon : +43 662 4470 24423
E-Mail : Andreas.KUTSCHE@spar.at
Internet: http://www.spar.at
------------------------------------------------
=
----LNX_Thu_Aug_12_2004_09:59:36_V3.33----