Performance-Issues after Update Kolab 3.4 to 16 on Centos7

[Roland Kolb]

Hello,

since I have updated my Kolab from 3.4 to 16 (Centos7) I have big 
performance problems.
It seems that this occurs mainly after users have added / changed 
entries in the calender.
During this performace issue no access to send, to receive mails, to 
change, to look on the calendar. After 30min or longer it is possible again.

When I check the memory (smem -u -t -k -w) with focus on apache, kolab, 
and cyrus I noticed that the numbers at the apache is very when the 
performance issue occurs.

During the normal status smem looks like

smem -u -t -k -w

User     Count     Swap      USS      PSS      RSS
libstoragemgmt     1        0   204.0K   207.0K   824.0K
rtkit        1        0   276.0K   291.0K     1.3M
rpc          1        0   580.0K   583.0K     1.1M
chrony       1        0   576.0K   608.0K     1.9M
avahi        2        0   500.0K   720.0K     2.1M
dbus         1        0     1.3M     1.4M     2.7M
colord       1        0     4.0M     4.3M     8.0M
postfix      5        0     6.9M     7.4M    22.3M
polkitd      1        0    23.1M    23.2M    26.7M
mongodb      1        0    52.9M    52.9M    54.0M
nobody       2        0   102.7M   102.9M   108.1M
nx           2        0   105.4M   109.0M   116.1M
gdm         26        0   198.2M   214.3M   314.0M
amavis       3        0   148.4M   220.0M   352.7M
cyrus      148        0   246.7M   248.8M   768.5M
kolab       29        0   288.5M   402.2M     1.2G
root        43        0   504.7M   551.8M   717.4M
mysql        2        0   597.3M   597.6M   600.8M
apache      24        0   715.8M   729.4M     1.1G
---------------------------------------------------


When the performance issues occurs it looks like (the other numbers are 
nearly the same)

cyrus       97        0   142.1M   144.4M   473.7M
kolab       29        0   179.0M   246.0M     1.3G
apache      54        0     1.5G     1.5G     2.6G


In the error message I found often looks like

stream_socket_client(): Could not connect to localhost:143: Connection 
refused in /usr/share/roundcubemail/program/lib/Roundcube/rcube_imap.php

or

kolab guam: 10:07:00.886 [error] gen_fsm <0.28450.1> in state 
passthrough terminated with reason: no function clause matching 
eimap:passthrough(process_command_queue, 
{state,"127.0.0.1",9993,true,true,{sslsocket,{gen_tcp,#Port<0.36308>,tls_connection,undefined},<0.2$
kolab guam: 10:07:00.886 [error] CRASH REPORT Process <0.28450.1> with 0 
neighbours exited with reason: no function clause matching 
eimap:passthrough(process_command_queue, 
{state,"127.0.0.1",9993,true,true,{sslsocket,{gen_tcp,#Port<0.36308>,tls_connection,undefined$

or in iRony/errors:
Failed to write to kolab cache
DB Error: Duplicate entry for key "PRIMARY" (SQL-Query: Insert into 
kolab_cache_event

or in maillog:
kolab amavis[3721]: (03721-09) prolong_timer 
ask_daemon_internal_connect: timer 10, was 282, deadline in 472.9 s
kolab amavis[3721]: (03721-09) ClamAV-clamd: Connecting to socket 
/var/spool/amavisd/clamd.sock, retry #2
kolab amavis[3721]: (03721-09) new socket by IO::Socket::UNIX to 
/var/spool/amavisd/clamd.sock, timeout set to 10
kolab amavis[3721]: (03721-09) (!)connect to 
/var/spool/amavisd/clamd.sock failed, attempt #1: Can't connect to a 
UNIX socket /var/spool/amavisd/clamd.sock: Datei oder Verzeichnis nicht 
gefunden
kolab amavis[3721]: (03721-09) get_deadline ask_daemon_internal - 
deadline in 472.9 s, set to 284.000 s
kolab amavis[3721]: (03721-09) prolong_timer ask_daemon_internal: timer 
284, was 10, deadline in 472.9 s
kolab amavis[3721]: (03721-09) get_deadline run_av_5 - deadline in 472.9 
s, set to 284.000 s
kolab amavis[3721]: (03721-09) prolong_timer run_av_5: timer 284, was 
284, deadline in 472.9 s
kolab amavis[3721]: (03721-09) (!)ClamAV-clamd av-scanner FAILED: run_av 
error: Too many retries to talk to /var/spool/amavisd/clamd.sock (All 
attempts (1) failed connecting to /var/spool/amavisd/clamd.sock) at 
(eval 130) line 613.\n
kolab amavis[3721]: (03721-09) (!)WARN: all primary virus scanners 
failed, considering backups


Has anybody the same issue? What was / were the problem(s)? Which tasks 
rises up the numbers at the apache? Only kolab runs on this server.

Thanks

Roland