Le 16/11/2011 14:22, Georg Bretschneider a écrit :
Le 14/11/2011 20:44, Jordan Clark a écrit :
As far as I can tell you have everything on one openvz container (MTA, LDAP, DB, IMAP, Web server, SOGo). Is
that correct? Are you running out of memory? I assume you are because you just updated the memory of the
container from 6 to 8GB. Compare your "maxheld" vs "barrier" of your
"oomguarpages" in your /proc/beancounters. If so do you know what is eating up your all of memory?
As for scaling to 3000+ users. You shouldn't probably have an issue with that.
SOGo probably won't be the bottle neck in the system as long as you set a
reasonable SOGo worker count. You really need to monitory the needs of all the
different parts of your mail system, find the bottle necks and resolve those
issues.
hope that helps,
Jordan
Hi Jordan,
i need to be more precise on our architecture
around SOGo:
LDAP: one physical machine
IMAP: one Xen virtual machine
MTA: one physical machine
SOGo itself:
one Container OpenVZ 8Go RAM, doesn't seem to be enough.
But as well, could you tell me more about your own SOGo environment
(potential number of users) even if you deployement isn't steal finish.
1) Did you take a look into your .GNUstepDefaults? Whats the value for
WOWorkersCount?
2) Whats the PREFORK value in your /etc/default/sogo?
You could aswell do
ps -aux | grep sogo
and look for the value behind -WOWorkersCount.
Could you provide us with this information?
Bye
Georg
Log reading with glogg http://glogg.bonnefon.org
<http://glogg.bonnefon.org/>
SOGoDebugRequest = YES;
Nb of child of sogod : 128
Nov 25 06:30:14 sogod [9683]: version 1.3.9 (build @vizzini.inverse.ca
201110281404) -- starting
Nov 25 06:30:14 sogod [9683]: vmem size check enabled: shutting down
app when *vmem > 384 MB*
Nov 25 06:30:14 sogod [9683]: <0x0x1ebd17a0[SOGoProductLoader]> SOGo
products loaded from '/usr/lib/GNUstep/SOGo':
Nov 25 06:30:14 sogod [9683]: <0x0x1ebd17a0[SOGoProductLoader]>
Appointments.SOGo, MailerUI.SOGo, SchedulerUI.SOGo, MainUI.SOGo,
ContactsUI.SOGo, CommonUI.SOGo, AdministrationUI.SOGo,
PreferencesUI.SOGo, Mailer.SOGo, Contacts.SOGo, MailPartViewers.SOGo
Nov 25 06:30:14 sogod [9683]: <0x0x1ebfe7b0[WOWatchDog]> watchdog
process pid: 9683
Nov 25 06:30:14 sogod [9683]: <0x0x2b3c11ab6b00[WOWatchDogChild]>
watchdog request timeout set to 10 minutes
Nov 25 06:30:14 sogod [9683]: <0x0x1ebfe7b0[WOWatchDog]>*preparing 128
children*
Nov 25 06:30:14 sogod [9683]: <0x0x1ebfe7b0[WOWatchDog]> child spawned
with pid 9692
[...]
Nov 25 06:31:11 sogod [9942]: <0x0x1ef5e430[SOGoCache]> Cache cleanup
interval set every 300.000000 seconds
Nov 25 06:31:11 sogod [9942]: <0x0x1ef5e430[SOGoCache]> Using host(s)
'localhost' as server(s)
WARN:
SOGoWebDAVAclManager ~100 to 650 entries /24h
UIxEnvelopeAddressFormatter ~ *44000 entries /24h*
Nov 25 06:31:11 sogod [9942]: [WARN]
<0x0x1f037ce0[SOGoWebDAVAclManager]> entry '{DAV:}write' already
exists in DAV permissions table
Nov 25 06:31:11 sogod [9942]: [WARN]
<0x0x1f037ce0[SOGoWebDAVAclManager]> entry '{DAV:}write-properties'
already exists in DAV permissions table
Nov 25 06:31:11 sogod [9942]: [WARN]
<0x0x1f037ce0[SOGoWebDAVAclManager]> entry '{DAV:}write-content'
already exists in DAV permissions table
Nov 25 06:33:13 sogod [9917]: [WARN]
<0x0x1f396060[UIxEnvelopeAddressFormatter]> unexpected envelope
address: <0x0x1f5e5560[NGImap4EnvelopeAddress]:>
these are only WARNings but it raised to 10% of the logs lines a day!
ERROR
~ 7000 entries /24h : "message: ATTENTION: utilisation non standard de"
2011-11-25 06:32:11.834 sogod[9942]
<0x0x1ec9a690[PostgreSQL72Channel]:
connection=<0x0x1ebeab10[PGConnection]: connection=0x0x1ec927c0>>:
message: ATTENTION: utilisation non standard de \\ dans une chaîne
littérale
always followed by these 4 lines
LINE 1: UPDATE sogo_user_profile SET *c_settings =* *'{"Mail"*:
{"Dr...
^
HINT: Utilisez la syntaxe de chaîne d'échappement pour les
antislashs, c'est-Ã -dire E'\\'.
The ^ symbol is under the *'* between *c_settings =* and *{"Mail"*
*Solution:
http://www.sogo.nu/fr/support/faq/article/i-get-nonstandard-use-of-%5C%5C-in-a-string-literal-in-my-sogo-logs-is-this-normal.html
*Could it be solved?
then i notice 2 types of information:
"successful login": 3000 up to 4000 entries /24h, is it good to
evaluate the day load?
Nov 28 09:47:26 sogod [3638]: SOGoRootPage successful login for user
'etu_user_id' - expire = -1 grace = -1
I'm interesting in your own count! Should be revelant to compare
installations.
Failure "Resource temporarily unavailable" : ~ 6 to 112 entries / 24h
during overload time
the 25/11/2011 6 times
the 28/11/2011, 18 times in the same hour [14h] (9 times at 14h42 and 6
at 14h44), then during [20h] 19 times
the 29/11/2011, (collapsing?) 29 times during [16h] (7 times at 16h08,
8 times at 16h16 and 11 times at 16h17)
the 01/12/2011, 31 times all day long
the 02/12/2011, 55 times all day long
the 05/12/2011, 90 times , top average of 10 times per hour
(11h-12h-13h-14h)
the week-end seems to be safe.
Nov 28 09:53:38 sogod [3510]: [ERROR] <0x0x4c6f0e0[WOWatchDogChild]>
FAILURE receiving status for child 3627
Nov 28 09:53:38 sogod [3510]: [ERROR] <0x0x4c6f0e0[WOWatchDogChild]>
socket: <*NGActiveSocket*[0x0x4b81b30]: mode=rw address=(nil)
connectedTo=<0x0x4b52590[*NGLocalSocketAddress*]:
/tmp/_ngsocket_0xdb6_0x49e7c70_000> receive-timeout=1.000s>
Nov 28 09:53:38 sogod [3510]: [ERROR] <0x0x4c6f0e0[WOWatchDogChild]>
exception: <*NGSocketTimedOutException*: 0x4b9e570>
NAME:*NGSocketTimedOutException* REASON:the socket was shutdown
INFO:{errno = 11; error = "Resource temporarily unavailable"; stream =
"{object = <*NGActiveSocket*[0x0x4b81b30]: mode=rw address=(nil)
connectedTo=<0x0x4b52590[*NGLocalSocketAddress*]:
/tmp/_*ngsocket*_0xdb6_0x49e7c70_000> receive-timeout=1.000s>;}"; }
Nov 28 09:53:38 sogod [3510]: <0x0x4c6f0e0[WOWatchDogChild]> sending
terminate signal to pid 3627
Nov 28 09:53:38 sogod [3510]: [ERROR] <0x0x4c6f0e0[WOWatchDogChild]>
FAILURE notifying child 3627
Nov 28 09:53:38 sogod [3510]: <0x0x4c6f0e0[WOWatchDogChild]> sending
terminate signal to pid 3627
Already discussed there:
https://inverse.ca/sogo/lists/arc/users/2011-10/msg00413.html
How to identify the origin of those sockets? (maybe unix(local),ldap,
imap...)
"fatal" entries: 15 to 30 /24h
Nov 25 08:38:14 sogod [9931]: <0x0x1f26dc20[GCSFolder]>
ERROR(-[GCSFolder writeContent:toName:baseVersion:]): cannot insert
content : <PostgreSQL72Exception: 0x1f26a360>
NAME:PostgreSQL72FatalError REASON:fatal pgsql error
(channel=<0x0x1ee91a20[PostgreSQL72Channel]:
connection=<0x0x1efc2860[PGConnection]: connection=0x0x1ee91af0>>):
ERREUR: valeur trop longue pour le type character varying(255)
is it important?
could it be fix?
"did not find locale for language" ~ 5000 entries /24h
Nov 25 06:38:47 sogod [9921]: [ERROR] [we-rm] did not find locale for
language: Romanian
Nov 25 06:43:21 sogod [9921]: [ERROR] [we-rm] did not find locale for
language: Chinese
Nov 25 07:19:25 sogod [9930]: [ERROR] [we-rm] did not find locale for
language: NorwegianBokmaal
Nov 25 07:27:49 sogod [9932]: [ERROR] [we-rm] did not find locale for
language: Finnish
Nov 25 08:05:22 sogod [9932]: [ERROR] [we-rm] did not find locale for
language: Slovenian
Nov 25 08:20:00 sogod [9932]: [ERROR] [we-rm] did not find locale for
language: Korean
Nov 25 10:58:47 sogod [9938]: [ERROR] [we-rm] did not find locale for
language: Slovak
Nov 25 11:36:57 sogod [22200]: [ERROR] [we-rm] did not find locale for
language: Japanese
Nov 25 12:09:03 sogod [22202]: [ERROR] [we-rm] did not find locale for
language: Danish
Nov 25 21:20:24 sogod [13382]: [ERROR] [we-rm] did not find locale for
language: Greek
is it possible to remove the choice for the Root Page : YES :
SOGoSupportedLanguages
but why those languages seems to be unavailable?
and why are they prompting?
Best practices to solved this?
I asked for a tool help me to analysed log.
grep and sort aren't enough cause it doesn't work with sequences!
Thanks all,
--
Stéphane GAUDICHE
--
[email protected]
https://inverse.ca/sogo/lists
