18 Sep
2007
18 Sep
'07
7:46 p.m.
Hello!
I wanted to upgrade a server from 1.1.0 to 1.2.2 today but despite
extensive testing beforehand i have some trouble:
After starting with the new openser-binary, it didn't allow any
registrations and answered with "500 server error occured".
Logging said:
Sep 18 17:23:12 [/sbin/openser] ERROR: sip_msg_cloner: cannot allocate \
memory_
Sep 18 17:23:12 [/sbin/openser] ERROR: new_t: out of mem:_
Sep 18 17:23:12 [/sbin/openser] ERROR: t_newtran: new_t failed_Sep 18
and
17:23:12 [/sbin/openser] ERROR: sip_msg_cloner: cannot allocate memory_
Sep 18 17:23:12 [/sbin/openser] ERROR: store_reply: can't alloc' clone \
memory_
after restarting it with "-m 128"(according to the faq) these errors are
gone, but it still doesn't accept registrations and prints:
Sep 18 17:24:16 [/sbin/openser] ERROR:tm:print_uac_request: no shmem_
Sep 18 17:24:16 [/sbin/openser] ERROR:tm:t_forward_nonack: failure to \
add branches_
Sep 18 17:24:16 [/sbin/openser] ERROR:tm:w_t_relay: t_forward_nonack \
failed_
[update] - after several restarts, the first error-messages appeared
again.
When i run the new version on port 5061(old production server 1.1.0 is
running parallel on port 5060) i can register with the new server(on
port 5061) without problems.
I have the vague impression that the new instance can't free memory,
but why does it work when i run it on a different port? Any clues?
Regards
Christian
19 Sep
19 Sep
9:17 a.m.
Are you sure that all processes of the old openser process are killed?
This looks indeed very strange.
klaus
Benko schrieb:
Hello!
I wanted to upgrade a server from 1.1.0 to 1.2.2 today but despite
extensive testing beforehand i have some trouble:
After starting with the new openser-binary, it didn't allow any
registrations and answered with "500 server error occured".
Logging said:
Sep 18 17:23:12 [/sbin/openser] ERROR: sip_msg_cloner: cannot allocate \
memory_
Sep 18 17:23:12 [/sbin/openser] ERROR: new_t: out of mem:_
Sep 18 17:23:12 [/sbin/openser] ERROR: t_newtran: new_t failed_Sep 18
and
17:23:12 [/sbin/openser] ERROR: sip_msg_cloner: cannot allocate memory_
Sep 18 17:23:12 [/sbin/openser] ERROR: store_reply: can't alloc' clone \
memory_
after restarting it with "-m 128"(according to the faq) these errors are
gone, but it still doesn't accept registrations and prints:
Sep 18 17:24:16 [/sbin/openser] ERROR:tm:print_uac_request: no shmem_
Sep 18 17:24:16 [/sbin/openser] ERROR:tm:t_forward_nonack: failure to \
add branches_
Sep 18 17:24:16 [/sbin/openser] ERROR:tm:w_t_relay: t_forward_nonack \
failed_
[update] - after several restarts, the first error-messages appeared
again.
When i run the new version on port 5061(old production server 1.1.0 is
running parallel on port 5060) i can register with the new server(on
port 5061) without problems.
I have the vague impression that the new instance can't free memory,
but why does it work when i run it on a different port? Any clues?
Regards
Christian
_______________________________________________
Users mailing list
Users(a)openser.org
http://openser.org/cgi-bin/mailman/listinfo/users
10:38 a.m.
On Wed, 19 Sep 2007 08:17:00 +0200
Klaus Darilion <klaus.mailinglists(a)pernau.at> wrote:
Are you sure that all processes of the old openser
process are killed?
This looks indeed very strange.
Yes, definitely - if i can trust `ps ax|grep openser` ;-)
netstat also shows nothing listening on port 5060 after killing the
old version and before starting the new one(I also remove /tmp/fifo if
it isn't deleted automatically).
I guess it would be gone when i restart the host but this is not
possible...
I also have no problems to start the 1.1.0-version again after killing
the new one...
regards
Christian
5:20 p.m.
I've recompiled openser with memory-debug now, as recommended in
http://openser.org/dokuwiki/doku.php/troubleshooting:memory but i
looks ok i think, i don't see any "not-freed memory"-entries(After
letting the server "rest" for 40 minutes without any traffic). Guess
i'll have to prepare a reboot if there're no other ideas...
0(32377) -----------------------------
0(32377) Memory status (shm):
0(32377) qm_status (0xb3ae2000):
0(32377) heap size= 67108864
0(32377) used= 16, used+overhead=23008, free=67085856
0(32377) max used (+overhead)= 1633712
0(32377) dumping all alloc'ed. fragments:
0(32377) 0. N address=0xb3ae71a8 frag=0xb3ae71a0 size=16 used=1
0(32377) dumping free list stats :
0(32377) hash= 1. fragments no.: 53, unused: 0
bucket size: 16 - 16 (first 16)
0(32377) hash= 2. fragments no.: 49, unused: 0
bucket size: 32 - 32 (first 32)
0(32377) hash= 3. fragments no.: 9, unused: 0
bucket size: 48 - 48 (first 48)
0(32377) hash= 4. fragments no.: 1, unused: 0
bucket size: 64 - 64 (first 64)
0(32377) hash= 6. fragments no.: 1, unused: 0
bucket size: 96 - 96 (first 96)
0(32377) hash= 8. fragments no.: 1, unused: 0
bucket size: 128 - 128 (first 128)
0(32377) hash= 32. fragments no.: 2, unused: 0
bucket size: 512 - 512 (first 512)
0(32377) hash= 38. fragments no.: 2, unused: 0
bucket size: 608 - 608 (first 608)
0(32377) hash= 65. fragments no.: 1, unused: 0
bucket size: 1040 - 1040 (first 1040)
0(32377) hash= 72. fragments no.: 1, unused: 0
bucket size: 1152 - 1152 (first 1152)
0(32377) hash= 97. fragments no.: 1, unused: 0
bucket size: 1552 - 1552 (first 1552)
0(32377) hash= 129. fragments no.: 1, unused: 0
bucket size: 2064 - 2064 (first 2064)
0(32377) hash= 143. fragments no.: 1, unused: 0
bucket size: 2288 - 2288 (first 2288)
0(32377) hash= 146. fragments no.: 1, unused: 0
bucket size: 2336 - 2336 (first 2336)
0(32377) hash= 225. fragments no.: 1, unused: 0
bucket size: 3600 - 3600 (first 3600)
0(32377) hash= 256. fragments no.: 2, unused: 0
bucket size: 4096 - 4096 (first 4096)
0(32377) hash= 640. fragments no.: 1, unused: 0
bucket size: 10240 - 10240 (first 10240)
0(32377) hash= 1031. fragments no.: 1, unused: 0
bucket size: 1048576 - 2097152 (first 1572864)
0(32377) hash= 1036. fragments no.: 1, unused: 0
bucket size: 33554432 - 67108864 (first 65475152)
0(32377) -----------------------------
0(32377) shm_mem_destroy
0(32377) destroying the shared memory lock
0(32377) Thank you for flying openser
6278
days inactive
6279
days old
3 comments
2 participants
participants (2)
-
Benko -
Klaus Darilion