Hi,
I cannot know by the logs what is causing the memory lick, it could
be caused by Appia, but also by other parts of Sequoia. These
messages are the cause of the high memory usage, because messages
start failing to appear on time and Appia starts to suspect that the
other controller has failed. So, these messages seem to be caused by
the memory lick.
There is nothing more in the logs that is not normal, like messages
being sent, but not being delivered, or something like that?
kind regards,
Nuno Carvalho
On May 24, 2007, at 6:45 , [EMAIL PROTECTED] wrote:
Hi @all,
we've a Scenario with 2 Controllers (Sequoia 2.10.6) and about 50
vdbs.
When I left yesterday, they both had a memory usage around 500MB.
When I arrived this morning, one of them used the complete RAM
(1024 MB).
(There is still the problem of the group communication, mentioned
in the
other topic).
There was no database usage the whole night (I know this, because
nobody
else but me do have any access to the databases.), why do the
memory usage
might grow like this?
The Logfile do not give a hint, only this interesting thing:
2007-05-23 14:42:11,716 INFO controller.virtualdatabase.test_big_50_C
First controller in group test_big_50_C
2007-05-23 14:42:11,719 INFO sequoia.controller.recoverylog Checking
recovery log consistency
2007-05-23 14:42:11,724 INFO controller.core.Controller Adding
VirtualDatabase test_big_50_C
2007-05-24 04:02:28,861 DEBUG continuent.hedera.gms Appia reported
block()
2007-05-24 04:02:28,987 DEBUG continuent.hedera.gms Appia reported
block()
2007-05-24 04:02:28,995 DEBUG continuent.hedera.gms Appia reported
block()
2007-05-24 04:02:29,508 DEBUG continuent.hedera.gms Appia reported
block()
2007-05-24 04:02:29,558 DEBUG continuent.hedera.gms Appia reported
block()
2007-05-24 04:02:34,830 DEBUG continuent.hedera.gms Appia reported
block()
2007-05-24 04:02:34,833 DEBUG continuent.hedera.gms Appia reported
block()
2007-05-24 04:02:36,251 DEBUG continuent.hedera.gms Appia reported
block()
2007-05-24 04:02:44,975 DEBUG continuent.hedera.gms Appia reported
block()
2007-05-24 04:02:45,266 DEBUG continuent.hedera.gms Appia reported
block()
2007-05-24 04:02:57,933 DEBUG continuent.hedera.gms Appia reported
block()
2007-05-24 04:02:57,936 DEBUG continuent.hedera.gms Appia reported
block()
2007-05-24 04:02:58,029 DEBUG continuent.hedera.gms Appia reported
block()
2007-05-24 04:02:58,044 DEBUG continuent.hedera.gms Appia reported
block()
2007-05-24 04:02:58,049 DEBUG continuent.hedera.gms Appia reported
block()
2007-05-24 04:02:58,056 DEBUG continuent.hedera.gms Appia reported
block()
2007-05-24 04:02:58,095 DEBUG continuent.hedera.gms Appia reported
block()
2007-05-24 04:02:58,098 DEBUG continuent.hedera.gms Appia reported
block()
2007-05-24 04:02:58,139 DEBUG continuent.hedera.gms Appia reported
block()
2007-05-24 04:02:58,196 DEBUG continuent.hedera.gms Appia reported
block()
2007-05-24 04:02:58,293 DEBUG continuent.hedera.gms Appia reported
new view:
version: 1
group: [Group:test_big_34_C]
id: [ViewID:2;[AppiaHedera@/xxx.234.68.193:36941]]
previous: [[ViewID:1;[AppiaHedera@/xxx.234.68.193:36941]],]
view: [[AppiaHedera@/xxx.234.68.193:36941],]
addresses: [/xxx.234.68.193:36941,]
2007-05-24 04:02:58,295 INFO continuent.hedera.gms
Member(address=/xxx.234.68.194:33472, uid=xxx.234.68.194:33472)
failed in
Group(gid=test_big_34_C)
2007-05-24 04:02:58,398 WARN controller.virtualdatabase.test_big_34_C
Controller Member(address=/xxx.234.68.194:33472, uid=xxx.
234.68.194:33472)
has left the cluster.
2007-05-24 04:02:58,408 INFO
controller.virtualdatabase.test_big_34_C 0
requests were waiting responses from Member(address=/xxx.
234.68.194:33472,
uid=xxx.234.68.194:33472)
2007-05-24 04:02:58,816 DEBUG continuent.hedera.gms Appia reported
new view:
version: 1
There follow a few more "faild in Group" entries, I did not check
until
now, but I think nearly all my vdbs left the cluster over night...
So there was nothing on the cluster for about 13 hours, and then
without
any reason, all the vdbs leave my cluster?! Might this be a reason
for the
high memory usage? Or could the memory usage be a reason for the vdbs
leaving the cluster?
Do anyone have an Idea for me?
_______________________________________________
Sequoia mailing list
[email protected]
https://forge.continuent.org/mailman/listinfo/sequoia
--
Nuno Carvalho
University of Lisbon, Portugal
http://dialnp.di.fc.ul.pt
_______________________________________________
Sequoia mailing list
[email protected]
https://forge.continuent.org/mailman/listinfo/sequoia
