Zabbix 5.0.36

Any news?

Hi,

Perhaps the following information will be useful. The problem occurs on the Zabbix environment version 5.0 and on the freshly deployed 6.0 - vSphere Client version 7.0.3.01400 - Hypervisor: VMware ESXi, 7.0.3, 21930508

Following the experiments, we came to the fact that the problem occurs when more than ~100 LUNs from the datastore are issued to the host.

Now the question is whether the problem with the limitation is on the side of VMware or on the side of Zabbix.
We will be very grateful for your check and reply

Kind regards,

Bogdan

Hi akurowski,

I have just tested with vc8.0.1

all works as expected.

Additional question:

why does your item use VMWARE.VM.UUID instead of VMWARE.HV.UUID

my item from template:

vmware.hv.hw.vendor[{$VMWARE.URL},{$VMWARE.HV.UUID}]

Hi Bogdan,

Be so kind as to provide more info about the problem:

1. which items are having a problem
2. what version of zabbix

Hi Michael Veksler,

Thanks fo reply.

1. All items and LLD got a issue message:
   The object 'vmodl.query.PropertyCollector:propertyCollector' has already been deleted or has not been completely created
2. Zabbix 5.0.36 on prod and Zabbix 6.0.17 on test, monitoring the same vCenter

As I wrote above, when we disconnect the datastore LUNs below 100, the data starts to read normally:

hence the question, is the module "vmodl.query.PropertyCollector" unable to process and collect more data, or maybe Zabbix has a limitation to download a larger data package?

Thanks,

B.

Bogdan - please provide the log of "vmware collector"

./zabbix_server -R log_level_increase="vmware collector"

I think the problem is the amount of hv info we are requesting. I'll try to find the request and reduce/split into smaller parts.

Sorry for the long response time, I was on vacation. Unfortunately, we cannot share logs with confidential data in a public place. Perhaps there is some other option to check?

Can you specify what "I think the problem is the amount of hv info we are requesting" means?
As I mentioned above, we have different esxi to which the same templates are attached, with the same number of items (as you can see in the screenshots, we tried different versions of the templates, hence sometimes it shows VMWARE.VM.UUID or VMWARE.HV.UUID). In addition, we have a freshly built Zabbix 6.0 test environment. We connected 2 esxi hosts there, one of which was collecting data, the other was showing an error. And we got the same result as in production. And, as I wrote earlier, we found a dependence on the number of connected LUN datastores. When more than 100 LUNs are connected on a host, for the whole host, all items get an error.

Kind regards,

Bogdan

Be so kind to report about your stopping point in Log by comparing it with this small ref.log reference (my log is from a master and there are a bit more steps there)

Hi @Bogdan Fortuna,
Be so kind as to help with questions:

• what is a max number of LUN per DS ?
• how many DS per 131 LUN ?

Hi MVekslers 

If you mean DS like a Data Store We have a 1 LUN per DataStore. Max number of LUN is 255 (256 is a number controller LUN of data matrix)

Hi @Bogdan Fortuna,

Interesting, as part of the ZBX-21153, I tested more than 120 ds per hv (v6.7).
Can you try testing monitoring through hv directly instead of vc ?

Sorry, old topic and I had to refresh my memory We had a problem when numbers of Data Stores was more than 100 per host/esxi, not LUNs:

Now, through migration, we have reduced the number of datastores per host and the problem has disappeared.

But, for example, I found the host with 183 LUNs connected:

and just 68 Datastores:

And data collect is work for this host. Than I think the problem occurs with more than 100 datastores connected to one host/esxi

Closing this for now, if you have any more questions, feel free to reopen this ticket!