NetXMS Support Forum

You do not need source node if you have configured it on NetXMS server node. It defines where actual data collection will happen, in case of "agent" origin - what agent will be queried. You'll need source node to be set in the opposite case - when you configure DCI on "myserver" node but SSH subagent is on NetXMS server - then you'll have to set source node to NetXMS server node.

Best regards,
Victor

Hi all!

We have updated server and agents to version 3.3.305. This version fixes some critical issues discovered in last few days:

- Server crash when receiving ill-formed SNMP response
- Server crash when accessing zone object attributes from NXS
- Error in initialization agent local database (issue NX-1849)

This version also contains driver for Moxa industrial routers.

Best regards,
Victor

First configure run fails because you didn't have "make" installed. I suggest to unpack source package into clean directory and re-run configure. You should not run automake and autoconf if you get source package - it may lead to broken configure. You should just run configure as it provided in the package.

Best regards,
Victor

Hi,

on this tab you will see charts for DCIs marked as "show on performance tab". It is configurable in DCI properties (right click on node, then "data collection", double click on any DCI).

Best regards,
Victor

Hi,

on what distribution you are building? How did you get sources - clone from git or source package?

Best regards,
Victor

Hi,

if you already migrated and dropped temporary tables you should clear pending background update indicator by executing query

UPDATE metadata SET var_value='' WHERE var_name='PendingOnlineUpgrades';

Best regards,
Victor

Hi,

yes, Ubuntu 20 packages will be available shortly.

Best regards,
Victor

No, this warning is only for TimescaleDB.

Best regards,
Victor

Currently there is no option for displaying node names on geo map.

Best regards,
Victor

If anyone experienced issues with unavailable login screen in web UI - it should be fixed in version 3.3.283, now available for download.

Best regards,
Victor

Hi all!

New stable release of NetXMS - 3.3 - is officially released. Changes since previous release:

- Option for blacklisting network device drivers
- Option for switching SNMP probing during configuration poll into single OID per request mode
- Automatic agent certificate renewal
- Value ranges in NXSL switch/case statement
- Explicit call of library functions in NXSL without "use" statement
- New attributes "instanceColumns" and "instanceColumnIndexes" in NXSL class "Table"
- New methods "readWebServiceList" and "readWebServiceParameter" in NXSL class "Node"
- New methods "applyTemplate" and "removeTemplate" in NXSL class "DataCollectionTarget"
- New NXSL functions: GetThreadPoolNames, sqrt
- New NXSL class "Template" that represents template object
- NXSL arrays can be implicitly converted to strings
- Implemented instance discovery method "Agent Table"
- Implemented instance discovery method "Windows Performance Counters"
- Added option for discarding collected performance data when database writer queue grows beyond limit
- Using timestamptz instead of integer for timestamp column in data collection tables on TimescaleDB
- Fixed issues:
        NX-392 (Use object names from Windows performance counters for instance discovery)
        NX-713 (Scheduled DCI will not work with offline DCI functionality)
        NX-806 (NXSL: add method to lookup table row by instance)
        NX-807 (NXSL: add method to lookup instance column ID without iterating over Table::columns)
        NX-814 (NXSL: add new node attribute – isRemotelyManaged)
        NX-1362 (Pattern matching for command output in SSH.Command parameter)
        NX-1511 (Save and display user ID of maintenance mode initiator)
        NX-1699 (Allow user turn on "Prevent automatic SNMP configuration changes" when creating node)
        NX-1729 (L2 maps get broken after server restart - all icons get coordinates (0, 0))
        NX-1732 (Option for discarding collected performance data when database writer queue grows beyond limit)
        NX-1765 (HTML codes visible in cell when viewing Table DCI last values in web management console)
        NX-1771 (Add NXSL constants)
        NX-1788 (NXSL: add support for ranges in switch/case)
        NX-1793 (Out of range error)
        NX-1806 (Change order in which ping3.ndd is used during discovery or allow unloading/blacklist)
        NX-1807 (TimescaleDB continuous aggregates, retention policies, etc.)
        NX-1671 (Title from Object Tool's properties is not used anywhere)
        NX-1808 (Template may not appear in object tree on import)
        NX-1820 (Object deletion and negative reference count)
        NX-1824 (Physical links - access denied)
        NX-1827 (Maintenance mode filter in dashboard element "Status Map")
        NX-1828 (Windows agent returns error for parameters System.FirewallProduct.* if built-in Windows Firewall is used)
        NX-1830 (Some events sent by agent may be lost if multiple events sent almost simultaneously)
        NX-1831 (Template path in node data collection configuration is incorrect)
        NX-1834 (Add NXSL function "sqrt")
        NX-1836 (Access attributes of template in auto-bind script)
        NX-1840 (Ability to set communication option "this is address of remote management node" during node creation)
        NX-1842 (Generate event when housekeeper process has started and finished)

Important note for TimescaleDB users

Because of idata/tdata table structure changes (timestamp column type changed from integer to timestamptz) database upgrade procedure divided into two parts. During offline (usual) schema upgrade old tables renamed by adding prefix v33_5_ (for example, idata_sc_default becomes v33_5_idata_sc_default) and new empty tables are created. You can start server afterwards and it will be fully operational, but without historical data. After that you should run nxdbmgr online-upgrade to migrate old data to new tables. This process could be time consuming and potentially require twice as much disk space as database normally uses. You may fully or partially truncate historical data tables before upgrade if you prefer to do fast upgrade without keeping historical data.

Best regards,
Victor

Unfortunately Jython (Python implementation in Java) which is used as foundation for nxshell seems to be stuck on version 2.2. If they eventually release Python 3 version we definitely will update nxshell.

Best regards,
Victor

Hi,

The fact that they do not support dot1dBase affects FDB reading. In FDB dot1dTpFdbPort supposed to contain bridge port number from dot1dBase. It looks like they just put ifIndex there. This we will be able to fix with device driver for those switches. I think that interface rename can be handled by driver as well. I think I'll be able to create driver for them in one of 3.3 patch releases if you can provide me access to test device or send required SNMP walks. You can contact me via PM regarding access or dumps.

Best regards,
Victor

Hi,

I have identified the issue. Problem is in how Fortigate builds index for ipAddressTable. For example, for address 192.168.188.1 OID in ipAddressTable looks like this:

.1.3.6.1.2.1.4.34.1.3.1.192.168.188.1.33

where .1.3.6.1.2.1.4.34.1.3 is base OID, 1 is address type (ipv4 in that case), 192.168.188.1 is IP address, and 33 is interface index (which is also strange because IP MIB does not specify interface index here). But the problem is that generic driver expects IP address prefixed with length, like in this example (taken from H3C switch):

.1.3.6.1.2.1.4.34.1.3.2.16.32.1.4.112.223.51.0.1.0.0.0.0.0.0.0.1

Here 2 is address type (ipv6), 16 is address length, and 32.1.4.112.223.51.0.1.0.0.0.0.0.0.0.1 is an address.

So when driver reads interface list from Fortigate it gets correct addresses from ipAddrTable and then picks incorrect addresses from ipAddressTable (168.188.1.33 in this example, shifted by one byte because server skips the one expected to hold address length).

I don't know what format is correct and what is not. Most devices that I have encountered does not support ipAddressTable at all and those that do returns addresses with length prefix. If somebody has better knowledge on this subject - please comment. For now I have added additional checks when reading ipAddressTable for correct value in length field - this should fix issue for Fortigate devices. It will be included into upcoming 3.3 release.

Best regards,
Victor

Just checked - there was a fix in DB init script between 3.2.400 and 3.2.472 for exactly this issue. Completely forgot about it.

Best regards,
Victor