Performance low in sogo-webfrontend

rowihei · November 3, 2016, 4:52pm

NethServer Version: 6.8
Module: sogo2.3.10

Good evening;

Mailcontimgent:

If I start sogo-webfrontend it needs

~3 s to show input
~ 7s to show sent
~ 3s to show trash
~ 7s to show sent (again)
~ 10s to show email

This times I found inside the local LAN (1GBit-Switch and cable are OK).
Mailserver needs between 50…60% RAM and have 2 SSD (mirror) ,CPU 2x3GHz , DNS is OK
Is there anything I can do to perform faster ?
How can I delete user faxmaster (don’t need faxserver) ??

Thank you for every idea !

Stefano_Zamboni · November 3, 2016, 5:21pm

can you please translate in english the labels I see in german? thank you

GG_jr · November 3, 2016, 5:25pm

Here you are:

rowihei · November 3, 2016, 8:06pm

Excuse me - here the same Server, but in english:

This server works for one year.

alefattorini · November 4, 2016, 10:29am

Please paste the output of the following commands:
free
top (sorting by MEM, click just SHIFT+M)

rowihei · November 4, 2016, 11:49am

Hallo,
rebootet nethserver… and now 95% mem is used, yesterday it was between 50…63%:

there the data you need:
free
total used free shared buffers cached
Mem: 3697772 3375540 322232 3036 31072 2190632
-/+ buffers/cache: 1153836 2543936
Swap: 3702780 33612 3669168

Excuse me - I forgot;
today is Friday and on Friday there is full backup at 12:30 - Thatswhy more mem in use!

Extract from log file /var/log/backup-data.log:

2016-11-04 12:30:03 - START - Backup data started
2016-11-04 12:30:09 - STEP - pre-backup-done done
2016-11-04 12:47:26 - STEP - backup-data-duplicity done
2016-11-04 12:47:29 - SUCCESS - Backup done

Now 16:30 shows this:
total used free shared buffers cached
Mem: 3697772 3237124 460648 3096 56280 2020652
-/+ buffers/cache: 1160192 2537580
Swap: 3702780 86440 3616340

Nethserver-Dashboard shows 88% mem in use and 2,3% swap (on SSD)

Thanks to you !

rowihei · March 23, 2017, 7:14pm

Good evening - it’s hard again …

now we have changed the hardware:

CPU:AMD Athlon™ 64 X2 Dual Core Processor 3800+ (2x2GHz)
RAM: 8GB (2x4GB for AMD)
2xSSD128 as RAID1
1GB-LAN

… but the same performance:
After starting the nethserver6.8 the first 10 minutes good performance at sogo-webfrontend, but than it becomes slower and slower and it’s no fun to work. That’s the show INSIDE LOCAL NETWORK.
If I try to connect from outside it’s not very fast, but it’s better to work as inside the LAN.
Structure is:

Firewall IPFire with Green+Red+Blue+Orange
Nethserver in orange-LAN (DMZ)
DHCP/DNS-Server is Windows-Server2012

Is there any idea to find the brakes ???
Thanks for every tip …

stephdl · March 24, 2017, 8:52pm

https://wiki.nethserver.org/doku.php?id=sogo#fine_tune_sogo

rowihei · March 26, 2017, 7:47am

Good morning,

thank you for the link Stephane.
But this is not logical. If I open sogo webinterface from outside (from home I mean), INBOX open needs ~3 seconds. But if I open the same INBOX inside the firm it needs ~10 seconds or more.
There is a firewall rule: red port 443 (https) > IPofNethserver (in orange). Thats OK and fast.
But the path from green to Nethserver (in orange) is much slower. The reason can’t be the sogo-config.
What to do ?

davidep · March 26, 2017, 7:53am

Does your Firewall perform some kind of https filtering from LAN to DMZ?

Also DNS reverse resolution from DMZ of IP in LAN could slow it down…

Just my 2c

stephdl · March 26, 2017, 8:26am

you can have also a side effect with sogo about java-script, if you want to do a test, you should use the same computer at home and at the firm. Indeed,now all the work to display the page is not on the server-side, but on the computer side (the principle of JS). Of course you might have a firewall Issue, let’s go to investigate.

Stefano_Zamboni · March 27, 2017, 12:37pm

exactly… SOGo3 is an heavy web app and to make it work fast you have to use a fast pc with plenty of ram…

don’t use IE for any reason

the issue is not the server but the client

be sure you’re not using any kind of proxy on the lan side for SOGo pages

rowihei · May 10, 2017, 6:17am

Hallo…AGAIN,

Thank you for your ideas, but I think this are not the reasons. Here there is an excerpt of the sogo.log.
IS THIS NORMAL or is there an error in sogo-sync ??

Stefano_Zamboni · May 10, 2017, 6:20am

You’d post on sogo’s BT

stephdl · May 10, 2017, 7:35pm

hi @rowihei

Do you imagine a human is able to read your log ?

rowihei · May 19, 2017, 2:42pm

Sorry the copy-paste eliminates the carriage return … and there are long-long lines
and sogo-direct (users@sogo.nu)- is not to handle for me, I’m too silly…
try again …

Hallo,
've running SOgo 2.3.10 on Nethserver 6.9.

Mailserver in DMZ of iPFire-Firewall

13 Users - 3 in office - 10 outside with smartphones (iPhone, iPad, Blackberry, Android)
.

Sometimes the time of reaction is very slow - especially in office(?! the green LAN) :

Login Sogo ~ 10…60 sec.
Open INBOX ~ 10…120 sec,
Open a mail: 5…180 sec.

But sometimes - most in the evening - it’s "normal fast"
I tried:

Mailserver in GREEN LAN > same problem
Clamavd stop > same
Testing DNS, traceroute > no errors found
time sync… OK

But found in sogo.log every second many [ERROR] and sleeping 5/10 seconds.
Is this “normally” ?
What should I do ?

Excerpt from sogo.log ::

"…
May 19 15:42:53 sogod [11115]: <0x0x7fd9f2c5e668[SOGoActiveSyncDispatcher]> Sleeping 10 seconds while detecting changes in Ping…
May 19 15:42:55 sogod [2563]: [ERROR] <0x0x7fd9f0a63cc8[WOWatchDog]> No child available to handle incoming request!
May 19 15:42:56 sogod [2563]: [ERROR] <0x0x7fd9f0a63cc8[WOWatchDog]> No child available to handle incoming request!
May 19 15:42:56 sogod [11044]: <0x0x7fd9f2a0d448[SOGoActiveSyncDispatcher]> Sleeping 5 seconds while detecting changes in Ping…
May 19 15:42:57 sogod [11256]: <0x0x7fd9f1e4cff8[SOGoActiveSyncDispatcher]> Sleeping 10 seconds while detecting changes in Sync…
May 19 15:42:57 sogod [2563]: [ERROR] <0x0x7fd9f0a63cc8[WOWatchDog]> No child available to handle incoming request!
May 19 15:42:58 sogod [11115]: <0x0x7fd9f2c5e668[SOGoActiveSyncDispatcher]> Sleeping 5 seconds while detecting changes in Ping…
May 19 15:42:59 sogod [2563]: [ERROR] <0x0x7fd9f0a63cc8[WOWatchDog]> No child available to handle incoming request!
May 19 15:43:00 sogod [2563]: [ERROR] <0x0x7fd9f0a63cc8[WOWatchDog]> No child available to handle incoming request!
May 19 15:43:01 sogod [11044]: 217.232.100.183 “POST /SOGo/Microsoft-Server-ActiveSync?User=user4&DeviceId=PVGNBGLRTL7VL33IPC0G6J6PM8&DeviceType=iPad&Cmd=Ping HTTP/1.1” 200 13/0 10.517 - - 0
May 19 15:43:01 sogod [2563]: [ERROR] <0x0x7fd9f0a63cc8[WOWatchDog]> No child available to handle incoming request!
May 19 15:43:02 sogod [11044]: <0x0x7fd9f2b82d78[SOGoActiveSyncDispatcher]> Sleeping 10 seconds while detecting changes in Ping…
May 19 15:43:02 sogod [11256]: <0x0x7fd9f1e4cff8[SOGoActiveSyncDispatcher]> Sleeping 5 seconds while detecting changes in Sync…
May 19 15:43:02 sogod [2563]: [ERROR] <0x0x7fd9f0a63cc8[WOWatchDog]> No child available to handle incoming request!
May 19 15:43:03 sogod [11115]: 80.187.115.61 “POST /SOGo/Microsoft-Server-ActiveSync?User=user6&DeviceId=EPDD6JTGHL1EP681DN70CKMEKO&DeviceType=iPhone&Cmd=Ping HTTP/1.1” 200 13/0 10.528 - - 0
May 19 15:43:03 sogod [11115]: 80.187.123.95 “POST /SOGo/Microsoft-Server-ActiveSync?Cmd=Sync&User=user1&DeviceId=androidc1311503317&DeviceType=BlackBerry HTTP/1.1” 200 12/0 0.004 - - 0
May 19 15:43:05 sogod [11115]: <0x0x7fd9f2c78438[SOGoActiveSyncDispatcher]> Sleeping 10 seconds while detecting changes in Sync…
May 19 15:43:05 sogod [2563]: [ERROR] <0x0x7fd9f0a63cc8[WOWatchDog]> No child available to handle incoming request!
May 19 15:43:07 sogod [2563]: [ERROR] <0x0x7fd9f0a63cc8[WOWatchDog]> No child available to handle incoming request!
May 19 15:43:07 sogod [11044]: <0x0x7fd9f2b82d78[SOGoActiveSyncDispatcher]> Sleeping 5 seconds while detecting changes in Ping…
May 19 15:43:07 sogod [11256]: <0x0x7fd9f1e4cff8[SOGoActiveSyncDispatcher]> Sleeping 10 seconds while detecting changes in Sync…
May 19 15:43:08 sogod [2563]: [ERROR] <0x0x7fd9f0a63cc8[WOWatchDog]> No child available to handle incoming request!

…"

and so on

Greetings and
thanks for every idea …

alefattorini · May 24, 2017, 9:58am

Are you using activesync devices? You could try to increase WOWorkersCount

config setprop sogod WOWorkersCount 15
config setprop sogod SOGoMaximumPingInterval 3540
config setprop sogod SOGoMaximumSyncInterval 3540
config setprop sogod SOGoInternalSyncInterval 30
signal-event nethserver-sogo-update

rowihei · May 24, 2017, 11:27am

Thank you, alefattorini ,
… it’s a miracle … yesterday I found in internet a way to improve performance

vim /etc/sysconfig/sogo
PREFORK=10
USER=sogo
vim /etc/rc.d/init.d/sogod
PREFORK=10

This changed the WOWorkersCount to 10
Now there are no more “[ERROR]…No child available to handle incoming request!” in sogo.log
Only many “Sleeping 10/5 seconds”

Now it’s fast enough. But I will wait a week to look again…
I will watch the RAM-usage
at 08:00 2711MB 00 swap
at 12:50 3201MB 00 swap

I don’t hope, that the next update re-change my improvement …
i think, the reason for slow performance are 10…12 smartphones in PUSH-mode.
After that, I will set to SOLVED (with a long way…)
Till next time …

stephdl · May 25, 2017, 10:47am

The next nethserver-sogo and sogo updates will overwrite your changes. You should follow the alessio’s tips

rowihei · July 2, 2017, 5:01pm

Hallo, again,

thanks to stephdl !
Your tip solved my problem. Now sogo-webfrontend is really fast. And the best of all - after update-installation the sogo-config is the same as before.
I watched the system some weeks and it’s OK. The RAM-usage climbed from 2000 MB to 3500MB.
But after full-backup it climbed to 5000 MB + 41MB swap - machine with 8GB!. System don’t clear the RAM-usage - but this is another topic …

Thanks and good bye !