Release : 10.4.0
Component : Spectrum Core / SpectroSERVER
Below are the fixes that are included in this BMP patch.
Symptom: Discovery console shows vEdge devices as "Unknown" when discovery process uses multiple SNMP community strings.
Resolution: Discovery console discovers the vEdge devices when the console consists of multiple SNMP community strings.
Symptom: Locator Search returns an error when using a list containing more than 250 items.
Resolution: Locator Search works as expected after refining search criteria filtering at OneClick server.
Symptom: SpectroSERVER crashes when a trap is received on 127.0.0.1 ip address (netsnmp).
Resolution: Fixed the SpectroSERVER crash issue when a trap is received on 127.0.0.1 ip address (netsnmp).
Symptom: Device model shows the connection to the LACP interfaces even they are removed from the device configuration resulting link aggregation alarms in Spectrum.
Resolution: False LACP alarms are not shown after the removal of the LACP interfaces from the device configuration.
Symptom: When LDAP server gets changed behind a Load-Balancer, it causes LDAP authentication issue and users cannot log into Oneclick Client.
Resolution: LDAP authentication issues are resolved when LDAP server is changed behind a Load-Balancer.
Symptom: Spectrum-CAPC Sync has poor performance.
Resolution: Improved Spectrum-CAPC Sync performance.
Symptom: OneClick Webapp takes more time to launch when connected through NAT VPN.
Resolution: Fixed the launch issue and now OneClick Webapp launches quickly when connected through NAT VPN.
Symptom: SpectroSERVER and SDC crashing with varying stacks rarely.
Resolution: Resolved the occasional multi-threaded memory corruption in core Spectrum code.
Symptom: OneClick is having high CPU issue with huge number of SLA Path models.
Resolution: SLA Path models are removed from SDN Manager hierarchy and provided through Locator search.
Symptom: Interface tab rendering is taking long time.
Resolution: Interface tab loads in less time while populating interfaces hierarchy.
Symptom: ArchMgr is crashing occasionally.
Resolution: Fixed the crashing issues of ArchMgr.
Symptom: Special chars in the event cause increase in the ArchMgr memory with flat file configuration.
Resolution: ArchMgr memory does not increase with flat file configuration when any special characters in the events.
Symptom: After device reconfiguration with new Interfaces, the LACP configuration is not correct.
Resolution: Device reconfiguration with the new Interfaces shows the correct LACP configuration.
Symptom: SpectroSERVER crashes when reconfiguring JnprPhyHub devices.
Resolution: Fixed the SpectroSERVER crash issues when reconfiguring JnprPhyHub devices.
Symptom: Device Search in OneClick with unrelated 9MB text results in CPU spike and performance mini dumps created at Spectro server.
Resolution: Device Search in OneClick with unrelated 9MB text results "Model Not Found" response without CPU spike and does not create any performance mini dumps at Spectro server.
Symptom: Old vpn sites with the no Interfaces still exists in the SSdb.
Resolution: The vpn sites with no interfaces are deleted from the SSdb.
Symptom: Aruba Access Point models are switching between Primary and Secondary controllers frequently.
Resolution: Only actively connected Aruba Access Points are modeled and associated to the controller. Switching is avoided until failover happens.
Symptom: Spectrum alerts flashing in/out periodically due to the following entry in the .vnmrc file orb_timeout=120
Resolution: Improved SpectroServer Heartbeat to prevent retry when SS startup. OC will retry SpectroServer Heartbeat in case of Heartbeat lost due to network issues, but not during the SS startup time.
Symptom: Logmon Feature is working good. Customer need special fix to handle one of their usecase where they can raise alarm on systemedge itself with the given formate.
Resolution: We have made changes to ensure we can raise the alarm of systemEdge itself once it is passed domain name is passed following the date and time as per the defined formate.
Symptom: ModelType BrocadeNos sets attribute modelclass to SNMP .
Resolution: The fix sets the model class of BrocadeNos devices as switches. Earlier the model class was improperly shown as SNMP.
Symptom: Watch status error text garbled.
Resolution: When evaluting the watch experssion returning junk error message which has hanlded. Now it will show as user defined error for error message.
Symptom: On Landscape disconnect and reconnect, OC looses all mpls manager LSP heads except the one of the reconnected landscape.
Resolution: On Landscape disconnect and reconnect, OC will preserve all mpls manager LSP heads of all the landscapes in DSS.
Symptom: MPLS Transport Manager Doesn't Model Nokia SR-OS devices.
Resolution: MPLS Transport Manager now models Nokia SR-OS devices.
Symptom: There is huge thread count increase.
Resolution: Threads are created only when the contact status is not equal to established, and threads wait toreceive a message only once.
Symptom: The connections are displayed in OneClick topology as black pipes in some cases.
Resolution: All the connections will be displayed correctly in OneClick topology and will no longer turn black.
Symptom: After reconfiguration duplicate stale modules still exist in the Interface Tab even after PortAgeOutTime expires.
Resolution: Duplicate stale modules will be removed once the PortAgeOutTime expires .
Symptom: Neighbors are not correctly displayed in the neighbors tab for WA_Link models.
Resolution: Neighbor models will be correctly displayed in the neighbors tab for WA_Link models.
Symptom: Informtaion Tab doesn't display any information with Spectrum 10.3.1. The Information Tab stays empty blank for ever for Cisco ASR routers which is having large IPRouteTable.
Resolution: Spectrum is now enhanced to read the scalar attribute rather than the list attribute.In View-devicedetails-config.xml there is check to display entry from which table based on ipRouteDest list attribute (0x100b1) . Since we are waiting to finish reading the attribute 0x100b1, we are facing this issue if it has lakhs of entries.There is a scalar attribute : ipCidrRouteNumber (id : 0x1292a)Instead of reading a list attribute, if we read a scalar attribute, the Information Tab will be displayed much faster. If the we are able to read 0x1292a attribute from the CiscoIPApp model then, we will display the table entries from the ipCidrRouteTable.If we are unable to read 0x1292a attribute, then we will display the table entries from ipRouteTable.
Symptom: SpectroSERVER performance problems, causing OC to red-box but not failover to secondary.
Resolution: Resolved several SpectroSERVER performance problems due to very high scale, and improved failover mechanism to properly failover to secondary when the primary is in busy/hung state.
Symptom: Issue while capturing host configuration where the last +-50 lines of the the config (startup or running) where not captured.
Resolution: Add below entries in .vnrmc and restart spectroserver./set ncm_tftp_wait=1/ncm_startup_wait=1//The values can be increased if not seeing improvement.// .
Symptom: Users can modify the watches using double click on the watch even if they do not have the privilege to do it.
Resolution: Users need privilege "Manage Watches" as part of the assigned role to modify the watches.
Symptom: Connections are not discovered between neighboring devices if LLDP versions are different on each of them (example one device with LLDPv1 and another device with LLDPv2).
Resolution: Connections are correctly discovered for neighboring devices with different versions of LLDP.
Symptom: PC Missing ItemPropertyTypeFlags for SdnId(it should be 26), and ds_item_member tables in CAPC database showing device and slapath model handles are same.
Resolution: ItemPropertyTypeFlags for SdnId is 26 and device model handle for slapath and tunnel are updated during spectrum sync to CAPC.
Symptom: DC Migration list showing empty and VCDCDomainList attribute of UIM Manager also is empty.
Resolution: 1. Code changes done to re-populate the DC list in next VM sync/2. Added locks to stop two syncs running same time and overiding the dc list. / .
Symptom: When VNA Integration enabled the models which are existed before integration are moving to VNA Container.
Resolution: When VNA integration enabled, the models which are existed before integration won't be moved to VNA Container.
Symptom: Float value for "percent" was not being accepted/.
Resolution: Change the datatype of "percent" to float.
Symptom: Black connections are observed between containers when there actually is no connection.
Resolution: Need to run the action 0x10900 on the VNM landscape handle to remove such black pipes.
Symptom: The xml does not comply to the expected grammar error message in VNM.out while searching for the models with Virtual entity unique ID.
Resolution: /Handled the error scenarios for xml grammer in Specrum UIM integration while searching for the models with Virtual entity unique ID.
Symptom: AP's modeled as SNMP shows the default values for SDN_OperationalStatus & SDN_AdminStatus.
Resolution: Access point's modeled as SNMP will show the Correct values of SDN_OperationalStatus & SDN_AdminStatus instead of default values.
Symptom: Report: Current Port Summary: Group is not setting the custom logo image in Report.
Resolution: Report: Current Port Summary: Group is set to use the custom logo image in the Report.
Symptom: Availability Group Report throws error when you have the common jar ca_customerMethods.jar along with spectrum_utils.jar in Jasper server.
Resolution: Availability Group report will work fine even if there is ca_customerMethods.jar present.
Symptom: Problem: NewMM.pl is not converting GnSNMPDev model type to AdvaDev.
Resolution: Core was generated while running the script and showed the memory access violation when we were trying the access the object which was deleted(m_saved).//Due to memory access violation when accessing the object, the conversion of the model type from one other failed. An updated script in this patch converts the device model type successfully.
Symptom: NCM bucket processing halted/failed with null pointer exception.
Resolution: NCM buckets gets processed without any exceptions or errors in tomcat log.
Symptom: Migrating wrong data center when searching for the data center from VMware data center modeling table and migrate./After migrating a data center of a vCenter, new data center which received in next sync is modeled under vCenter of migrated landscape.
Resolution: Migrating selected data center which is filtered from VMware datecenter modeling table for migration./After migrating a data center of a vCenter, new data center which received in next sync is modeled under vCenter original landscape.
Symptom: SNMP v3 Profiles not always retaining the Authentication Password.
Resolution: Split AuthPassword and PrivPassword only on the basis of first occurrence of "^"./This is done to support "^" character in passwords.
Symptom: When a site model does not exist or cannot be read from MIB, \u003CcodeVpnDiscovery.cc(1947): Interface name Rsp error ignoring 0\u003C/code was printed in VNM.out.
Resolution: The logging of the warning message in VNM.OUT is diabled now while creating the site model.
Symptom: Customer was observing additional connections created due to LACP.
Resolution: Make LACP connections discovery as configurable. By default the behavior is true i.e. LACP connections will be discovered by default.
Symptom: Error while running Jasper report "Current Ports Detail: All" due to component_OID .
Resolution: The report "Current Ports Detail: All" is fixed to treat component_OID as a string instead of float.
Symptom: In Legacy Spectrum UIM Integration, when UIM models are removed from group, VMware sync delta processing is not detecting the model removed from group and retaining it in inventory.
Resolution: In Legacy Spectrum UIM Integration, when UIM models are removed from group, VMware sync delta processing is detecting the model removed from group and moving the model to unreported inventory.
Symptom: Spectrum Incremental Sync with CAPC takes log time and eventually it fails/time-out. Spectrum Syncs unnecessary models as Tunnels/SLAPaths.
Resolution: Spectrum Incremental Sync with CAPC will be quicker and will not fail/time-out. Spectrum does not Sync unnecessary models to CAPC.
Symptom: No Alarm generated when Access point is down.
Resolution: Added check to prevent scheduling of two attributes NRM_DeviceCPUUtilization (0x12aaa)/NRM_DeviceMemoryUtilization (0x12ac6) for models derived from pingable.
Symptom: UIM integration sync takes a long time .
Resolution: Connection discovery is now done just once for all the devices, instead of sequentially per device. UIM integration sync performance is significantly faster.
Symptom: search for active v3 remote profile is taking time.
Resolution: removed active v3 remote profile set and check part.
Symptom: Setting SDC name in SNMPv3 profile was not supported.
Resolution: Added support for setting SDC name in SNMPv3 profile.
Symptom: The new fault codes are not supported in spectrum and unknown severity faults treating as a new alarms in spectrum.
Resolution: The new fault codes which are coming as a notification updates are supported in spectrum./The notification severity is unknown and state as modified of Notification object treating as clear alarm in spectrum. .
Symptom: device IP address was not filled in v3 traps for blink action.
Resolution: Setting agent address in device IP address of trap object in v3 traps for blind action and Using Agent Address in filter when device ip filter is configured.
Symptom: Spectrum-10.4 oneclick.jnlp href generation changed. It is creating url href with static FQDN of the webserver.
Resolution: OneClick.jnlp href generation will use the Hostname from request(Host Header) instead of static FQDN of webserver.
Symptom: SRM asset report shows devices IP address for the interfaces when they are not configured with any IP.
Resolution: SRM asset reports shows empty IP address for interfaces and doesn't not show the device IP address.
Symptom: SpectroSERVER would crash while deleting the timer object of type MONOSTABLE.
Resolution: SpectroSERVER should NOT crash while deleting the timer object of type MONOSTABLE.
Symptom: Huge number of Management lost alarms seen on SNMP devices in oneClick client.
Resolution: No more Management lost alarms seen on SNMP devices in oneClick client.//Note: Changing the value of snmpEngineID on device has important side-effects(causes genuine MAL Alarms on device ). A user's password (entered on the command line) is converted to an MD5 or SHA security digest. This digest is based on both the password and the local engine ID. The command line password is then destroyed, as required by RFC 2274. Because of this deletion, if the local value of engineID changes, the security digests of SNMPv3 users will be invalid, and the users will have to be reconfigured on device./ After reconfiguration of engineID, Username and password on device, customer needs to execute update action(./update action=0x10333 mh=<VNM Handle>) on VNM model via VnmSh for removing previous cached engineID,boot time, etc values from SpectroSERVER so that SS will retrieve the new engineID values from device during next polling/discovery process.
Symptom: Spectrum 10.3.2 SS is crashing when integrated with SDC 10.4.1.
Resolution: SdmSDManager::processMessage method loggin the SDC message, which have the "uid" value has null pointer and causing the crash.Problem occurs only if debug is enabled and SS is 10.3.2 or below and SDC is 10.4.1.10.4.1 SDC is sending NCM_CONNECTION_STATUS_UP and NCM_CONNECTION_STATUS_DOWN messages to SS.10.4.1 SS code is handling these messages in SdmEtpkiEndpoint::read() method, but10.3.2 code is not handling this code, hence SdmSDManager::processMessage method being called andSdmMessageContext object is setting "uid" attribute to null.in debug logging, stringstream c++ object appending with this value. It is crashing in windows but linux this is not crashing(may be in linux stringstream is handling null values).
Symptom: Error while running Service Summary by Service Name Jasper report when security is enabled in SRM preferences. .
Resolution: Report query is corrected for Service Summary by Service Name.
Symptom: Getting wrong digest report from device every time.
Resolution: Calculating authentication and privacy keys again and re-sending packet in case of wrong digest report.
Symptom: Float value for "percent" was not being accepted.
Resolution: Change the datatype of "percent" to float.
Symptom: Periodically, OneClick will freeze. It happens most often when viewing a Topology, Interfaces tab and Information tab.
Resolution: Upgrading to latest Java version 220.127.116.11 resolves the freezing of OneClick client.
Symptom: Error while running the Group Reports when multi-tenancy enabled.
Resolution: Report query is fixed for Group reports.
Symptom: connected SDC(10.4.) to MLS primary and Secondary with -loglevel debug enable/then upgraded sdc to 10.4.1 (then done couple of sanity things like sdc stop and start and ncm services restart )//sdc was up and no crash is seen at that time//it may got crashed after 10 or 15 mins after that.
Resolution: UID value coming as null and so added the null check to avoid the crash.
The excessive logging we were seeing related to BST alarms ( tomcat log was over 5GB) - 10.4.1.
You may contact Spectrum support to request this patch.