UCS 4.3/4.4 Domain and Management Console: Modules not loaded in most cases


#1

Hi there,

since the very first install of UCS, the web-based domain and management console does not load the chosen module when clicking on a module.

Sometimes it works and the module is loaded. Most time it fails. The problem occurs with almost all modules, like DNS, DHCP, Users. The problem occurs even when accessing the web-based domain and management console from different computers and different web browsers. Sometimes, the “Users” module does not load, whereas the “DNS” module loads. Sometimes, it is exactly vice verse. I access UCS usually from different Mac computers. I got the impression that the likelihood of getting a module loaded is higher when working with the Safari browser. But also there, I have a success rate of about 10% to get the module loaded. I need to try again and again, wait, and suddenly, the module loads and shows its content.

In most cases, clicking on a module opens a new page with the module name in the headline without showing any content in the window. This remains unchanged. Sometimes, I receive an error message:

Ein unbekannter Fehler mit Status-Code 502 trat während des Verbindungsaufbaus zum Server auf. Bitte versuchen Sie es später noch einmal.

However, in most cases, no response is shown. When I try 10-20 times, suddenly the module loads and I can work.

Does anybody has a hint to overcome this behavior? What debug information could I check to identify the source of this problem?

Your comments are welcome. Thank you very much.

Best regards,
Peter


#2

Hello @herrep,

HTTP Status Code 502 is “Bad Gateway”. This might be a problem in the network setup. I assume the error is shown via JavaScript in your browser (maybe you can share a screenshot of the error message?) and probably not caused by something in the UCS backend. Can you explain the network layout? Which path should network packets take from your browser to UCS and back again, which network devices need to be passed (firewalls, proxy, routers …)?

Best regards,
Michael


#3

Dear @Grandjean,

Thank you very much for your reply. The full error message (in German) is:

Ein unbekannter Fehler mit Status-Code 502 trat während des Verbindungsaufbaus zum Server auf. Bitte versuchen Sie es später noch einmal.

However, this error message does not appear often and only after a long waiting time, like a timeout. In most cases, simply nothing happens, i.e. the module page opens with the header containing the module name but without displaying any content.

The problem also occurs within my local network where UCS is installed as a VM on a Synology NAS so that there is one virtual network interface in between. Even if I use a client that is connected at the same switch like the Synology NAS, the timeout issues ocur, but less often as compared to a connection via Internet.

I wonder what debugging information should be retrieved to identify the problem.

Best regards,
Peter


#4

Hi,

I do not have a real solution for this issue but I noticed it happens more frequently when the UCS server does not get enough hardware power or is very busy with other tasks.

So your Synology might not be able to offer enough resources to the UCS-VM,

Could you monitor your underlying hardware to see if there is some sort of bottleneck?

/CV


#5

How much memory does your VM have? How much swap? And hoch much is used? = Please post the output of free


#6
root@server:~# free
              total        used        free      shared  buff/cache   available
Mem:        6115112     3956844      298060       47004     1860208     1840780
Swap:       4194300     1856860     2337440


#7
top - 09:34:35 up 32 days, 10:06,  2 users,  load average: 2,44, 2,28, 1,77
Tasks: 197 total,   2 running, 194 sleeping,   0 stopped,   1 zombie
%Cpu(s): 57,2 us,  2,8 sy,  0,0 ni,  0,0 id, 38,9 wa,  0,0 hi,  0,5 si,  0,7 st
KiB Mem :  6115112 total,   241992 free,  4031104 used,  1842016 buff/cache
KiB Swap:  4194300 total,  2393068 free,  1801232 used.  1766868 avail Mem 


#8

Yeah, that server has way too little memory available. You should size a server’s memory so that it doesn’t have to swap during regular usage and still have at least 1.5 GB of memory available for buffers & cache. In you case your server needs to swap out 1.8 GB of memory, resulting not only in slower operation but likely in out-of-memory scenarios — and the symptoms you described sound a lot like out-of-memory situations.

I’m curious what you’re running on your server that takes up that much memory. Please post the output of ps auxw --sort=vsz


#9

Hi,

this is quite a high load (2,44) but it might depend on the count of CPUs you have assigned to the UCS. For two or one CPU it is a relative high load, for more CPUs it should be fine.
What bothers me is the high 38,9% of IO-Waits together with the very high and variable usage of the swap. I bet this is (at least) one reason for the behaviour.

Please chek this article regarding performance.

/CV


#10

I run a Nextcloud server which currently performs occ fulltextsearch:index over the entire files. The corresponding PHP process already occupies almost 2 GB RAM. Once this process is finished, I will add further 2 GB RAM to the VM.


#11

At present, 2 CPU cores are assigned. The host resources should also allow 4 CPU cores.


#12

Just to inform… using hp server dl360 gen5 (Intel® Xeon® CPU E5450 @ 3.00GHz )running xenserver i cannot have any ucs vm running… i have the same issues until i give up…

Server has 32 gb i give the vm 12 then 16 gb of ram all the cpus etc but always have the same issue…

In the similar server but with less memoy i can run it as physical server… i never can understand what was the issue