CMS Idea Exchange

Currently it is required to log into a VCenter system to create the running software CI to kick of the VMWare jobs.  Customers are now using many instances of VMWare Appliances (no longin capability except via Web UI on port 443) which host their VCenter servers. We would like to add another job to discover these VCenter Appliances that would then be able to create the running software CI needed by the other VMWare jobs. ... View more

A job will complete but will show warning status. This is misleading because there ARE errors in the job, even though it shows only a warning status.  (submitted on behalf of Cigna) ... View more

By default the "IBM Virtualization by Shell" job discovers both active and inactive LPARs and there is no way to change this behavior. The discovered LPARs' state (Running, Not Activated) is shown in the "LPar state" attribute of the related "Lpar Profile" CI.                                         It is a common practice for administrators to disable but not delete virtual systems after they have been decommissioned, or create temporary systems, templates etc. which are inactive. The result is that all these are being discovered,  consume licenses, and are not deleted by the uCMDB aging mechanism, as they are touched by the IBM Virtualization job. So a true/false discovery parameter is needed in order to choose whether or not inactive LPars will be discovered. There is a similar parameter in "VMware vCenter Topology by VIM" job, namely "reportPoweredOffVMs". The same should be implemented for "IBM Virtualization by Shell" job.   ... View more

Our end customer ΟΤΕ is planning to upgrade its IBM HMCs to version 9. These are consoles used for monitoring the IBM Virtualization infrastructure. Since the latest supported version of IBM HMCs in "IBM Virtualization by Shell" job is version 8, this is either a stopper for upgrading, or there is a risk that they will not be able to properly discover their crucial IBM Virtualization infrastructure after the upgrade. So it is suggested that IBM HMC version 9 is officially supported in a future CP update. ... View more

Currently any DB query which takes longer than 1000ms will be also logged in cmdb.dal.slow.log This is not always relevant as a lot of queries are expected to run for more than 1 second. this is a hardcoded value. We should be able to have the OOTB value higher than 1000ms and this value should be controlled by a setting so not hardcoded. ... View more

Normalization rules are very useful as they alter the result vector before being sent to the server on each probe. The resut vector is enriched/normalized on each probe in a descentralized manner so the server receives the desired form or the result vector. The workaround currently used is to alter the CI data from discovery with the help of enrichments. This can have server side performance and that data will pass thorugh the reconciliation engine twice: 1st time comming from discovery and the 2nd time after being altered by an enrichment. This causes a lot of overhead and can be avoid with normalization rules if we have the case of a simple triplet. The restrictions of the normalization engine are clear in http://cmshelpcenter.saas.hp.com/CMS/10.22/ucmdb-docs/docs/eng/doc_lib/Content/dfm/adap_config_c_rules_syntax.htm for 10.22 Points in this rule which are not supported now by the current normalization engine and should be available: example, servername is a custom attribute <ns2:normalization-rule ci-type="file_system" id="9666">             <rule-input>                 <attribute name="\x00" value="servername" compare-type="regexp" />             </rule-input>             <rule-output>                 <attribute name="servername">                     <value>$root_container</value>                 </attribute>             </rule-output>         </ns2:normalization-rule> We can’t use as rule-input an attribute which isn’t in the result vector. Here we use servername attribute which is never modelled via Jython scripts so it will never be in the result vector of discovery. We can’t use a regular expression for a null value of an attribute We can’t assign in the rule-output a parameterized value to an attribute. In our example, we can’t clone an attribute value on the same attribute. We can’t reuse attributes in any form (we can use only static values) The question would be what is the impact on probe side. ... View more

When running discovery jobs that have agents installed, sometimes the build teams will not follow the advice of the UCMDB team and they will just clone all the agents to new systems.  This causes a rash of cloned UD_Unique_ID values which, in turn, cause massive reconciliation overhead and creates 'supernodes' in UCMDB that contain all the information from all affected CIs. I would request we add the following to VERIFICATION rules in node identification: <verification-criterion> <attribute-condition attributeName="ud_unique_id" operator="EqualIgnoreCase" includeNullValue="false" conditionType="approveAndContradict"/> <attribute-condition attributeName="name" operator="EqualIgnoreCase" includeNullValue="false" conditionType="approveAndContradict"/> </verification-criterion> This would have the effect of NOT merging CIs based only on a UD Unique ID being the same IF the names are different.  Then the teams can track which UD Unique IDs have been duplicated and they can help the build teams to remediate that without impacting data quality inside UCMDB.  In addition, since the validation rule exists still for UD Unique ID, the call home events would be processed normally. ... View more

Hello, we have a requirement to discover the Dynatrace-Appmon environment; specifically the "System profiles" that define the monitoring specifics for the various infrastructure and application components. Is there a plan to include this capability in a future discovery module? Regards, Benoit ... View more

Would like to add Special Monitoring of the SOLR/Zookeeper environment. When Zookeeper goes down the uCMDB Browser search will indicate that no results are found even if they exist. Also if both SOLR instances go down then we will be in the same boat. The Browser should give the message .. Search in not working SOLR Search Engine is down contact administrator...  (Submitted on behalf of Cigna) ... View more

We need to be able to analyze information that comes to us from the JMX console in a form that doesn't require manual data manipuation.  We should be able to pipe JMX output to a form that can be read by other tools.  For example, be able to output the JMX run like viewJobStatuses into a CSV format. (submitted on behalf of Cigna) ... View more

Looking for better error messages to help troubleshoot issues.    When a <GC overhead limit exceeded> message is triggered by the application it should be displayed into the communication log so we can fix the root cause of the problem faster. <Submitted on behalf of CIGNA> ... View more

It is often necessary to inform user about status and change of UCMDB server. This ER would enrich of actual Splash screeen fuctionalty. Functionality requested: -broadcast a message to all user (for example for an upcoming maintenance operation) -when UCMDB server is down, URL should display an outage page   ... View more

Deployment of content packs does not always go as expected or a feature works differently than expected.  The ability to uninstall a content pack and restore the system to its previous state would be very helpful and would reduce the risk and effort associated with implementing new CPs (submitted on behalf of Cigna) ... View more

Find the Primary IP address using the Host connection by shell job. This would be in: Universal Discovery - Content - Host Connections This will be especially helpful for Asset Manager integrations.  In  AMPushAdapter, TCPIPAddress in AM is mapped with PrimaryIPAddress attribute in uCMDB. This could be an alternative to the Inventory Scanner.   Inventory by scanner is missing to populate for a lot of machines the hwNICIPAddressFlags attribute.  It's missing from the scan file and thus the Primary_ip attribute won't include any IP. This could also help in clustered environments, for example issues with changing attribute for some runningSW CIs which are linked to some nodes part of clustered environment.  The primary_IP attribute is a good indicator of what is running on what IP address, it can make the difference between the clustered IP and the IP of the host which is part of the cluster.  Some customers can't use Inventory on their env due to different restrictions and those customers will benefit from the proper primary_ip attribute. (Submitted on behalf of Cigna) ... View more

We need be able to create relation between two existing CIs by theirs ucmdbIds. This new feature should be implemented for next APIs - Rest API, JAVA API and especially for Federation API (we are using implementation of PopulationAdapter interface).   Explain in details (I will use Rest API for the example below). Rest API has Update CI method, user able to update any CI by it's ucmdbId like this - calling PUT method   http://servername:port/rest-api/dataModel/ci/4fc5cb9b413954a397f10ee777a305f8 with body     { properties: { data_note: sometext } } Data note will be updated for CI with the UcmdbId.   But Rest API doesn't have method to create relation between two existing CIs by their ucmdbIds.  At this moment Rest API (and others) cannot do such simple operation, to do this user have to use Insert Topology method, but this method requires building a new CI like this:   { cis: [{ ucmdbId: [temp_id], type: [ucmdb_cit], properties: { <ucmdb_cit_prop>: [ucmdb_cit_prop_value], … } }, …], relations: [{ ucmdbId: [temp_id], type: [ucmdb_cit], end1Id: [temp_id], end2Id: [temp_id], properties: { <ucmdb_cit_prop>: [ucmdb_cit_prop_value], … } }, …] }     PROBLEM - In real world and real business cases external system will never have all information required for Reconciliation Data-In, especially when we are talking about node CI type and it's components. In fact in most cases external system knows UcmdbIds and a little bit another information. In our business case we need to integrate UCMDB and Microsoft SCOM, but in MSSCOM all required information don't exist, so we can't use Insert Topology method at all. ... View more

We have deployed the UD Agent pretty much across the board on our estate. I know we could use SSH connections, but it would be great if we could be consistent with our discovery of Linux instances on the IBM mainframes.   ... View more

This is a request to simplify the process of generating new UD Unique ID. - A short description of the present behaviour The generation of a new ud_unique_id is not documented well and not user friendly. Currently when you want to generate a new unigue id (uda) on unix you need to: - Go through unclear documentation - Use 2 different commands (with confusing names discagent and discagnt) - Switch between different folders - Steps to reproduce the behaviour/issue Example for Linux systems (but is relevant for all UNIXes) 1) Log in using the user account that is used to run the UD Agent (e.g. root) 2) Navigate to the directory that contains the UD Agent executable files (for Linux    /opt/microfocus/Discovery) 3) Stop UDA agent ./discagent stop 4) Switch to folder /opt/microfocus/Discovery/bin 5) Generate a new unigue id by running  ./discagnt --newuniqueid 5) Navigate back to the directory  /opt/microfocus/Discovery 6) Start the agent by running  ./discagent start - The behaver you desire in a future release Simply enhance current script discagent to support parameter newid Extend the current shell command  discagent with new parameter to perform newunigueid operation. Now the shell command supports only parameters start, stop, restart and status. Add a new option for example newid So the simple command ./discagent newid  would generate a new unique id (it would stop the agent, generate new unigue id, start the agent). Generally discagent should be the only "wrapper/handler" to perform all actions with UDA agent for administrators.  - Impact on the production Simplify the support of cloning and imagining. Better UDA management. Less confusing. Faster to adopt by administrators in UNIX world. - Whether you can’t wait for a future release and require a hotfix We can wait for the next release - Business justification if a hotfix is requested Hot -fix is not required   Expectations: 1. Simplify the steps, preferably consider single command rather than two 2. Document the clear steps in a guide ... View more

We would like a separate discovery job or an option to Vmware vCenter Topology by VIM job to only discover the ESX hosts and relationships. We have some vCenters that have lab or temporary build VMs that we are not interested in collecting any data about the VMs or their relationships. We are interested in discovering the underlying physical infrastructure that is the ESX Hosts. With newer versions of vCenter there is no longer an OS or option to access the ESX Hosts directly to discovery attributes such as Model, Serial Number, CPU, Memory and Host name. The only option is to use the Vmware VIM API to query the vCenter about the associated ESX Hosts. We would like to see from the vCenter host down to the ESX host including the relationship to the vCenter cluster and vCenter Datacenter folders. This information is needed in order to uniquely identify each ESX Hosts. There are cases were poor management of ESX Hosts will leave the host name localhost and the only difference between ESX Hosts is the parent folder structure. We would like to see this as a basic inventory job. ... View more

Memory, especially storage, should have its value in GB rather than MB (e.g. logicalvolume_size of CIT LogicalVolume, memory_limit of CIT VMware Host Resource, disk_size of CIT FileSystem). ... View more