Unable to delete hosts or unlink templates after discovery...

[Alfista]

Hi,

i have use discovery for finding hosts in my network in Zabbix Appliance 3.4.8. After that I have added the template Linux SNMPv2. And there started the problems. On some hosts the Template find about 6000 Items and on some normal count (about 200).
After that I tried to remove the template (unlink with clear data), then delete the host but nothing will work.

Please can you help me with these problems:
1. how to unlink the template with clearing the data (standard template unlink with clear data in host settins doesn't work), that I don't have there so many items and I don't overload the zabbix server?
2. how to delete host with so many items (standard delete in host tab doesn't work)?
3. can some one explore me why the Linux SNMPv2 Template - HOST-RESOURCES-MIB SNMPv2 has generate by discovery so many items on one linux system and on on other it find excat the actual count of items and how to limit it, that it doesn't make these problems? The problem is exactly in memory Discovery and vm.memory function. It found there abourt 1000 memories and added to each 4 items, then the CPU where it found more CPU's (cores), but this isnt so bad, while it looks that it generated items to HT pipelines.

Thanks for helps.

[aigars.kadikis]

Hi,

Are these 6000 items mainly in a supported state? Are you able to remove one of them manually? What Applications category has assigned for the most of these 6000 items?

3. There are three discovery rules in 'Template Module HOST-RESOURCES-MIB SNMPv2' - storage, memory, CPU. You need to work with Filter section and create some exclusion rules to decrease the item count in one of your servers.

After you set new exclusion filters then please set temporary 'keep lost resource period' to 0.
and decrease your discovery rule 'Update interval' (for example to 10m) just to take faster effect.

Can you clarify which discovery rule has generated so many items in your situation?

2. I hope the third point will solve the second one.

Regards,

[Alfista]

Hi Aigars,

yes 80% are fuly active and the others are calculated and some of them are unsuported while division by zero.
Yes im able to remove manualy item by item. On one I have done it and needed a half day for it :-(.
It has added the Template - HOST-RESOURCES-MIB SNMPv2, the Application Memory and teh Memory Discovery Rule and the Discovery items in it. All was preinstalled in Zabbix.

The filters are set as it was created in Zabbix:



Where should I set the 'keep lost resource period'?

Yes, as I wrote above it is the Memory Discovery rule. I dont know hy it has find so many memories, when in older version when I had installed it was OK and in this devices are not so many memories ...

Thanks.

[aigars.kadikis]

Hi Aigars,

yes 80% are fuly active and the others are calculated and some of them are unsuported while division by zero.
Yes im able to remove manualy item by item. On one I have done it and needed a half day for it :-(.
It has added the Template - HOST-RESOURCES-MIB SNMPv2, the Application Memory and teh Memory Discovery Rule and the Discovery items in it. All was preinstalled in Zabbix.

The filters are set as it was created in Zabbix:



Where should I set the 'keep lost resource period'?

Yes, as I wrote above it is the Memory Discovery rule. I dont know hy it has find so many memories, when in older version when I had installed it was OK and in this devices are not so many memories ...

Thanks.

'keep lost resource period' can be found under tab [Discovery Rule]. Please show some item key names which is in supported state and which you would like to not include in monitoring.
Please include the screen which contains error message 'division by zero'.

[Alfista]

Hi Aigars,

here are the screen shots:

[aigars.kadikis]

Hi Aigars,

here are the screen shots:

Please list the STORAGE_TYPE on the host where you have a lot of items:

snmpwalk -v 2c -c public localhost:161 .1.3.6.1.2.1.25.2.3.1

[Alfista]

Here is it.
It was needed to yip it, while it has 0,5MB.

It is a Linux system build on FreeBSD clang version 3.3 and on previous version of Zabbix 3.4.6 id doesn't create so many items.

[aigars.kadikis]

We can discover and include description in discovery. This can be done by replacing your discovery rule SNMP OID from

 discovery[{#SNMPVALUE},1.3.6.1.2.1.25.2.3.1.3,{#ALLOC_UNITS},1.3.6.1.2.1.25.2.3.1.4,{#STORAGE_TYPE},1.3.6.1.2.1.25.2.3.1.2]

to

discovery[{#SNMPVALUE},1.3.6.1.2.1.25.2.3.1.3,{#ALLOC_UNITS},1.3.6.1.2.1.25.2.3.1.4,{#STORAGE_TYPE},1.3.6.1.2.1.25.2.3.1.2,{#STORAGE_DESCR},1.3.6.1.2.1.25.2.3.1.3]

and choose the most important things in HOST-RESOURCES-MIB::hrStorageDescr like 'Isilon_2576' and 'rl_entry'


Regards,

[aigars.kadikis]

Sorry, but now i think the better way would be to work with hrStorageSize. Please test out this discovery:

discovery[{#SNMPVALUE},1.3.6.1.2.1.25.2.3.1.3,{#ALLOC_UNITS},1.3.6.1.2.1.25.2.3.1.4,{#STORAGE_TYPE},1.3.6.1.2.1.25.2.3.1.2,{#STORAGE_SIZE},.1.3.6.1.2.1.25.2.3.1.5]

and set regex as:

^[1-9][0-9]*$

Regards,

[Alfista]

OK and what it will do?
It will filter the received data to have only the needed and not all?
The 2nd filter fill show only info which doesn't have 0MB? And in the 1st it will be released to the Isilon in description?

I don't fully understand it. The filter based on size is OK but the 1st based on description not. There will not have all items description Isilon when they are from Isilon?

Thanks for exploration.

[aigars.kadikis]

OK and what it will do?
It will filter the received data to have only the needed and not all?
The 2nd filter fill show only info which doesn't have 0MB? And in the 1st it will be released to the Isilon in description?

I don't fully understand it. The filter based on size is OK but the 1st based on description not. There will not have all items description Isilon when they are from Isilon?

Thanks for exploration.

As you said the second filter will exclude all items which has StorageSize 0 (in other words: which is bigger than zero)
The first rule is based on description and yes - it not very usefull this time.

There are also such a thing like StorageUsed with numeric address .1.3.6.1.2.1.25.2.3.1.6
Please consider compare StorageUsed with 0. This would generate more items than in case of 'StorageSize'.

Discovery:

discovery[{#SNMPVALUE},1.3.6.1.2.1.25.2.3.1.3,{#ALLOC_UNITS},1.3.6.1.2.1.25.2.3.1.4,{#STORAGE_TYPE},1.3.6.1.2.1.25.2.3.1.2,{#STORAGE_USED},1.3.6.1.2.1.25.2.3.1.6]

Filter:

{#STORAGE_USED} = ^[1-9][0-9]*$

I believe one of those exclusions can eliminate message 'division by zero'.

Let us know how it goes.

Regards,

[Alfista]

Ok and the macro should be:
{#STORAGE_USED} or as above {#STORAGE_SIZE} ?

[aigars.kadikis]

Ok and the macro should be:
{#STORAGE_USED} or as above {#STORAGE_SIZE} ?

Like in discovery string mentioned we are going with {#STORAGE_USED} scenario.

Regards,

[Alfista]

OK changed. Will see what happens.

[Alfista]

Nothing changed. The result is the same. Added above 4000 items and also the with 0 too.

[aigars.kadikis]

Would you mind to share how you configured the discovery rule and the filter tab?

[Alfista]

Of course.

In the Discovery rule i have changed the SNMP OID as you sent me:

discovery[{#SNMPVALUE},1.3.6.1.2.1.25.2.3.1.3,{#ALLOC_UNITS},1.3.6.1.2.1.25.2.3.1.4,{#STORAGE_TYPE},1.3.6.1.2.1.25.2.3.1.2,{#STORAGE_USED},1.3.6.1.2.1.25.2.3.1.6]

And here are the screen shots:




I don't fully understand it. On other Linux PC it works correctly, only this one not.

[Alfista]

And I found that this last template doesn't show the founded memory storages as the old one, with the correct names, that it can be better identified.