DX Infrastructure Management

[askCA TRANSCRIPT] CA Unified Infrastructure Management – March 22, 2018 

Mar 26, 2018 01:38 PM

from Melissa Potvin (CA) to everyone: Hi Everyone, ok let's get started !

from Melissa Potvin (CA) to everyone: I have support engineers with me and product management ready for your questions.

from Praveen to everyone: Hi Team- What is the amount of transaction logs that UIM generates in daily basis, During a particular time UIM generates 40 – 50 % GB of TR logs where whole DB goes to non- responsive state. Any suggestions

from Gene Howard to everyone: @Praveen, Our best practice guid is to set sql server in imple mode so the transaction logs do not cause a performance issue.

from Gene Howard to everyone: @Praveen, there is no way to provide you a ball park on this number as it will depend on the number of QOS you are collecting and the how often you are collecting them.

from Gene Howard to everyone: @Praveen, we do have a database sizeing spread sheet we can provide but it will not tell you the size of the transaciton logs that will be generated.

from Melissa Potvin (CA) to everyone: if you just joined our askCA event, thank you. feel free to ask questions ! support engineers and product mgmt is here for you.

from Praveen to everyone: Yes Gene DB is in simple mode only . I have enabled mirroring and even it breaks due to heavy logs . How can i enable addtionallogging to find which query is causing huge logs duringa specfic time alone

from Gene Howard to everyone: @Praveen, what database and version are you running?

from Fernando Cozzo to everyone: I have heard the NETTRAFIC probe will no longer be available. Is this true?

from Mike Arnone to everyone: SOI has a database tool built-in to do various tasks, such as DB cleanup, etc... Are there any plans for UIM to have a DB Tool similar to SOI?

from Gene Howard to everyone: @Praveen, I would think this is being cause by nightly maitenance where we are summarizing data and deleteing data

from Praveen to everyone: @ Gene MS SQL 2012 Enterprise

from Lawrence Atlas (CA Technologies) to everyone: @Mike, Net_traffic is already end of support as of September 2015.

from Praveen to everyone: manitenance is scheduled every night , but i am facing this issue in moring time

from Gene Howard to everyone: @Praveen, does this seem to happen daily after midnight? do you get an error or something? I think we would need to work with your DBA on this help further.

from Fernando Cozzo to everyone: was there supposed to be a replacement or is it deprecated? Are we supposed to use Spectrum SNMP Objects to monitor a Server's Interface?

from Gene Howard to everyone: @Mike, currently the data_engine and nas do some database client up. We do not have a seperate utility for this. This would be a great question to bring up on the road map session on March 29th with details.

from Gene Howard to everyone: @Mike did you have something specific you were looking to have done that is not?

from Praveen to everyone: dataengine and its dependency probe is going down during that time DBA has confirmed it is due to heavy TR logs from UIM

from Lawrence Atlas (CA Technologies) to everyone: @Fernanado net_traffic was replaced by snmpcollector.

from Fernando Cozzo to everyone: SNMPCollector is a remote or local probe managed by the server controller?

from Fernando Cozzo to everyone: are we supposed to load MIBs somewhere for this now?

from Praveen to everyone: How do I get system up time report in days,hours ,min format .. Last time in same session I got info as it can be customized I missed to email can you suggest on this.

from Lawrence Atlas (CA Technologies) to everyone: SNMPColletor is a remote probe. https://docops.ca.com/ca-unified-infrastructure-management-probes/ga/en/alphabetical-probe-articles/snmpcollector-snmp-data-monitoring

from Gene Howard to everyone: @Praveen, in 9.0 there are some fixes for database connetivity issue where the data_engine will reconnect if there is some time the database is not avaiable... I would suggest if you have not to open a support case so we can get more details around the issue and engage dev as needed

from Praveen to everyone: Thanks Gene !!

from Tero to everyone: Any news about the 9.0 release date?

from Lawrence Atlas (CA Technologies) to everyone: For MIBS, you can look at these 2 websites to see if the device is supported .https://devicesupport.ca.com/ https://docops.ca.com/ca-unified-infrastructure-management-probes/ga/en/alphabetical-probe-articles/snmpcollector-snmp-data-monitoring/snmpcollector-snmp-data-monitoring-release-notes/update-vendor-certifications

from Gene Howard to everyone: @Tero cureently we are expecting end of Q2 2018 / End of June 2018

from Lawrence Atlas (CA Technologies) to everyone: If the device is not on either list, then you will need to open a support case for device certification. Device certification runs on a 3 month release cycle.

from Melissa Potvin (CA) to everyone: link to location where you can register for roadmap session: https://www.ca.com/us/product-roadmaps.html

from Gene Howard to everyone: Please regsiter for a road map session if you would like to know more about UIM 9.0

from Tero to everyone: Ok. Thank you

from Praveen to everyone: It is possible to change severity of robot inactive alert for some servers alone?

from Ben Nelson to everyone: @Praveen. As you're no doubt aware uptime is in seconds since last reboot. We do have a feature in the backlog to change how this is calcuated (by the hub) and we are also adding OOB availability reports in CABI. We are targeting this for UIM 9.1 but it may be earlier.

from Fernando Cozzo to everyone: how does this change how QOS metrics would be collected now if we have to use SNMP... it seems we need to move away from onitoring interfaces with UIM and transition to Spectrum/eHealth even on certain Virtual Guest servers

from Ben Nelson to everyone: UIM 9.0 is scheduled to release at the end of Q1 (June 2018)

from Lawrence Atlas (CA Technologies) to everyone: You can use snmpcollector to monitor interfaces through UIM.

from Gene Howard to everyone: @Praveen currnetly no this can not be changed, unless you use a nas script to change it.

from Ben Nelson to everyone: @Praveen. regarding robot inactive alarms. This is in the backlog and will be coupled with the up/down metric which we're adding for availibility reports.

from Praveen to everyone: Thats Great Ben!!

from Praveen to everyone: How to monitor server hung state, server responds to icmp and robot is responding and sending QOS how to overcome this.

from Gene Howard to everyone: @Praveen currently dev is looking into better ways to report on this. This kind of indeterminent state is very hard to diagnose and report on. As you mention somethings work and some do not and it can be different from instance to instances. If you have suggestion on how this might be able to be accomplished please do bring to the product team via and idea or the road map sessions

from Praveen to everyone: Sure Gene

from Praveen to everyone: I am monitoring SQL service in nt-services probe in Microsoft fail-over cluster server if the services is moved from one to another I am getting service down alerts in node 2 . Whether it will work as this or alert will get suppress as service is moved to node 2.

from Lawrence Atlas (CA Technologies) to everyone: What probe is generating the alarm?

from Praveen to everyone: nt-services

from Gene Howard to everyone: @Praveen are you using the cluster probe in this setup? If not you might want to look into that

from Gene Howard to everyone: this would move the profile from one cluster node to another and not alarm on the down cluster.

from Praveen to everyone: Yes gene i have cluster probe too

from Gene Howard to everyone: Short of that it will have to be a manual p;rocess.

from Gene Howard to everyone: did you setup the ntservices in the cluster probe?

from Melissa Potvin (CA) to everyone: Good for you Praveen for asking all your questions here! who else has questions, anyone?

from Gene Howard to everyone: if you did and it is alarming on the down then you need to open a support case to look into why

from Praveen to everyone: Whether i need add in service in profile section ..

from Praveen to everyone: I have raised support case ,but they confirmed it is not possible and i could see cluster probe supports ntservices probe.

from Gene Howard to everyone: so yes the need to have the profile setup in the shared section so the probe knows to move this between instances.

from Praveen to everyone: there is any offline docs to do this i could not find anthing in online docs

from Gene Howard to everyone: Do you have a case number? I can review if you like

from Praveen to everyone: pls share u r mail ID i will mail you gene right now i dont have

from Gene Howard to everyone: eugene.howard@ca.com

from Praveen to everyone: In IM robot is showing in red state, but there is no alert triggered, how to trouble shoot in this case.

from Praveen to everyone: There is no robot inactive alert*

from Gene Howard to everyone: @Praveen, Set the hub loglevel to 3 and logsize to 35000, then remove the nimsoft\hub\robots.sds file and restart the hub. you should see some information about the robot and communication

from Gene Howard to everyone: the robot inactive alert should be comging from the hub it is attached to. Would need to check the hub.log on loglevel 3 or higher and the nas log as well to make sure the alarm is not being supressed

from Praveen to everyone: Checked in nas there is no alerts genereated from hub itself .

from Gene Howard to everyone: Sorry no check the nas log on loglevel 3 or higher. there may be a pre-processin script supressing it. May want to use Dr nimbus as well to see if the alarm is being sent. if you are not already make sure you are using a recent hub version suich as 7.93

from Praveen to everyone: Sure i will rempve the robot.sds part which i didnt tried before.

from Praveen to everyone: It is possible to customize search options by IP address or by host name in UMP similar to Spectrum locator search .(load set of servers and get the output)

from Gene Howard to everyone: @Praveen are you talking something like the trend reports you can run on a group? Sorry not that familiar with spectrum....

from Tero to everyone: Any experience how many tunnels and queues a (unix) hub can handle? We are facing issues where our client hubs are starting to crash now and then

from Gene Howard to everyone: @Treo so usually on a linux box client start reporting issues when they have more than 40 tunnel connections. Clients that keep the number of tunnel clients below this seem to be fine.

from Praveen to everyone: @Gene i need a input box where i can enter the ip addrsess and export my results

from Gene Howard to everyone: @ Praveen what resutls, I would assuume QOS information

from Praveen to everyone: inventory info Gene

from Gene Howard to everyone: @ Praveen sorry not that I know of currently, this would need to go on our idea wall....

from Praveen to everyone: Thanks !!will submit idea

from Mike Arnone to everyone: Are there plans to make Net_Connect & RSP probes where it can be automated, rather than have to make all changes one at a time via the GIU ?

from Praveen to everyone: I want to generate report of Virtual machines which are connected with media like mounted iso ,as of now with probe it is not possible any work around can be done.

from Tero to everyone: @gene Thanks. We have 80 tunnel clients per tunnel hub and same amouint of active queues and ~90 subscribers. per hub. Might be issue then and we need more tunnel hubs to share the load

from Gene Howard to everyone: @ Praveen not without a custom probe or scripting. As QOS information has to be numeric and not strings,, for a report it would be hard to be anything but yes or no..

from Gene Howard to everyone: @Tero YES, I would think you need to spread the load out

from Melissa Potvin (CA) to everyone: 5 minute warning for final questions please.

from Ben Nelson to everyone: @Mike, MCS (Monitoring Configuration Service) goes a long way to address some of this. Both net_connect and rsp can have a monitoring prfile configured against a group. As you are no doubt aware USM groups are dynamic and when/if group membership changes are made, MCS will push out the new profiles (and probes for local probes). Does this cover what you mean by "automated"?

from Praveen to everyone: if I reset MCS ,where there will be any impact to probe and it configs which has done earlier

from Gene Howard to everyone: @Praveen what do you mean reset?

from Gene Howard to everyone: if you wipe out the tables then the probes and configues already deployed will stay in place until a setup in MCS is done again that overwrites them

from Praveen to everyone: I hope the configs stays in robot folder as probe .cfg files

from Melissa Potvin (CA) to everyone: OK, i believe we answered all questions. is everyone all set ? was this session useful?

from Praveen to everyone: i am reffering this KB Gene https://comm.support.ca.com/kb/how-to-reset-mcs-monitoring/kb000012764

from Tero to everyone: @Melissa yes. thank you

from Gene Howard to everyone: yes if you follow that you will wipe out the tables

Statistics
0 Favorited
3 Views
0 Files
0 Shares
0 Downloads

Related Entries and Links

No Related Resource entered.