Strange slowing performance

[vargacs]

Dear all,

My ZCS Open Source Edition in my company sometimes has strange function. It is random starting to slow once about 3-5 days. It is slowing down minute by minute, and becomes unavailable until I restart. And it is working fine for the next random slowing. The Webmail, admin console, ssh is available, but it is lose the connection with the server (or database I think). I saw the logs but is shows nothing.
I have about 650 user. I installed on Hyper-V High Availability Cluster. We have 4 Dell Poweredge R730XD with 2x12 core CPU, 196GB ram. For the zimbra I added 16 vCPU and 32GB ram. I thing it is more thang enough. My processor usage is never go up to 20% and the memory usage is about 5-6GB. The Storage is attached to host with 2x8GB Fibre Channel.

I see some diagram, and I found some strange value (i think). The strange datas only in mysql. I made some screenshot here: https://mega.nz/#F!PVNzmSLD!H4mTqWzAMJhk-Rh8VsXFrA
Every decline in the diagram is a restart.

Is there an error on my system? Can I handle this, or have I make a cron to restart it every evening?

Thanking in advande.
Csabi

[phoenix]

You should always give details of the ZCS version that's in use by posting the full output of the following command:

Code: Select all

zmcontrol -v

It's impossible to give you any advice on your problem as you've given no details of the workload on your system nor what type of users and how many at any one time, are they IMAP or fat client users, web users, what the mail load on this server, etc., etc., etc.....

As for the configuration I'd say you have far too many vCPUs allocated to this server, you need to allocate the minimum necessary not 'as many as you can'. You need to do some reading of the articles on ZCS performance on the wiki (there are plenty there, if you look) but start with this one: https://wiki.zimbra.com/wiki/Performanc ... re_vSphere

[vargacs]

You should always give details of the ZCS version that's in use by posting the full output of the following command:

Code: Select all

zmcontrol -v

It's impossible to give you any advice on your problem as you've given no details of the workload on your system nor what type of users and how many at any one time, are they IMAP or fat client users, web users, what the mail load on this server, etc., etc., etc.....

Sorry for that, you have right. My output is:

Code: Select all

zimbra@mail:~$ zmcontrol -v
Release 8.7.0.GA.1659.UBUNTU14.64 UBUNTU14_64 FOSS edition.

I have about 650 users, ~350-400 users at the same time only with IMAP, and webmail.

As for the configuration I'd say you have far too many vCPUs allocated to this server, you need to allocate the minimum necessary not 'as many as you can'. You need to do some reading of the articles on ZCS performance on the wiki (there are plenty there, if you look) but start with this one: https://wiki.zimbra.com/wiki/Performanc ... re_vSphere

Thanks for it, I will read it, and I will change the settings.

[vargacs]

As for the configuration I'd say you have far too many vCPUs allocated to this server, you need to allocate the minimum necessary not 'as many as you can'. You need to do some reading of the articles on ZCS performance on the wiki (there are plenty there, if you look) but start with this one: https://wiki.zimbra.com/wiki/Performanc ... re_vSphere

Now I dont understand.. Here a fresh blog about the installing on Virtual Environment. I followed this at install procedure: https://blog.zimbra.com/2016/05/install ... 14-04-lts/

2.- OS Requirements
In this Blog article, we are using Ubuntu 14.04 LTS on VMware vSphere 6.0U2. These are the Requirements for a Single-Server VM with up to 50 users:
4vCPU or more. I have 4x4vCPU
8GB de RAM or more. I have 4x8RAM
50GB could be the minimum recommended, remember to use the fastest available disks and adjust the size of the server according to your Environment
Internet connectivity This is a requirement in order to install Zimbra Collaboration 8.7, which uses external packages repository, and of course, you need Internet conectivity to send and receive emails
A perfect DNS configuration is needed

So i think the 16 vCPU and 32GB RAM for 400 active users at the same time is not too mutch. Am I wrong? If I take back the vCPUs to 8 what will be change in mysql load? (I will be happy if need only 8 vCPU and 16GB RAM.)
My Disks in RAID 60 array (24x600GB) 10.000rpm SAS Disks.

I dont understand.

Thanks,
Csabi

[vargacs]

Here is some Mail report, If it is helps.

Friday:

Grand Totals
------------
messages

6932 received
40025 delivered
0 forwarded
24 deferred (247 deferrals)
31 bounced
8843 rejected (18%)
0 reject warnings
0 held
0 discarded (0%)

756m bytes received
5023m bytes delivered
1886 senders
635 sending hosts/domains
1384 recipients
403 recipient hosts/domains

Thursday:

Grand Totals
------------
messages

6325 received
39080 delivered
0 forwarded
27 deferred (228 deferrals)
43 bounced
10664 rejected (21%)
0 reject warnings
0 held
0 discarded (0%)

930m bytes received
5146m bytes delivered
1652 senders
720 sending hosts/domains
1318 recipients
410 recipient hosts/domains

Wednesday:

Grand Totals
------------
messages

8050 received
44102 delivered
0 forwarded
48 deferred (220 deferrals)
33 bounced
9434 rejected (17%)
0 reject warnings
0 held
0 discarded (0%)

1050m bytes received
6251m bytes delivered
2399 senders
702 sending hosts/domains
1639 recipients
439 recipient hosts/domains

[vargacs]

So nobody can help me?

Tapatalkkal küldve az én SM-G930F eszközömről

[vargacs]

I found the solution for my problem. When I installed ZCS, the memory was only 8GB for the install, after I raise it to 32GB before we started to use global. I had to set in the my.conf the innodb_buffer_pool_size, innodb_log_file_size to the right size for load the database to the memory. The MySQL load is perfect now.

Thanks for the help!

Csabi