Highlighted
Outstanding Contributor.
Outstanding Contributor.
123 views

Seeing duplicate hosts - can we safely delete one without affecting metrics data?

Jump to solution

Due to some change in the environment (not sure what/when), after a restart of a Hyper-V proxy, two Hyper-V hosts that were previously reported with upper-case hostnames are now being reported with lower-case names.

This has resulted in CO seeing these as new hosts, and reporting that the old ones are no longer responding, generating the orange warning popup in the UI.  Running pvconfig -lt shows both instances of the host, along with the corresponding collection times/dates.

Running pvdump, however, does not appear to recognize the two distinct versions of these hosts, as all hosts are reported with short hostnames in upper case.

I want to use pvconfig -dt -ty HYPERV -t <target> to remove the invalid target name, but I am concerned about removing any associated data that has been collected against the "old" name.

Is this a concern?  Does pvconfig -dt remove metrics data, or just the collection configuration?

Thanks,
Aaron

Labels (1)
Tags (2)
0 Likes
1 Solution

Accepted Solutions
Highlighted
Micro Focus Expert
Micro Focus Expert

Re: Seeing duplicate hosts - can we safely delete one without affecting metrics data?

Jump to solution

Hello Aaron,

You bring up a good question.

Assuming that I understand correctly, it sounds like you have two Hyper-V hosts that are now showing twice within the list as targets even though nothing has changed within the proxy agent that was restarted.  Is this correct?

While my next example is not the same, I'm sharing it because it may make a difference.

I recently upgraded my KVM server from CentOS 7.5 to 7.7.  When I looked at my CO server today, I noticed that the KVM server was still listed as one target but the Host was listed twice when I used the pvdump command.  One entry that was current and the other that had not been updated since the date of the upgrade.  When I looked closer, I found that the SystemID entry had changed as part of the upgrade.

I bring this up simply that there may be any number of things that can cause an issue like this to occur.

Here is what I did to look more closely at the situation.

You can run the 'pvsupport.sh -grab' command to collect not only the logs but an export of the CO DB.

If you extract the resulting file on your workstation, you can go into the Support/Database/InstanceTables directory to find the inst_Infrastructure__Node.csv file.

What I normally do is open this within Excel, select the first column, use the 'Test to Columns' under the 'Data' menu, select 'Delimited' and add the "|" to the 'other' entry and then Finish.

This should make all of the entries in separate columns and then I use the column filters to isolate the entries that I'm looking for, etc.

Example:
Filter 'Role' for 'HOST'.
Filter 'HostName' for problem hosts.
Filter 'HostHostName' for problem hosts to see the guests associated with the hosts and the latest collection time.
Compare SystemID field if duplicate host entries exist.

I would look at the Guests that are part of those two problem hosts and confirm that they are showing as being updated and are not duplicated.

I know that this was not your question, but it is a way that I use to look at things more closely.

To answer your question, the 'pvconfig' command should only modify the target entries.  These entries are also located within the output of the 'ovconfget pvcd.hyperv'.

Ex.
# ovconfget pvcd.hyperv
CollectionType=REMOTE
Targets=DOMAIN\couser#hypv101v3.company.com:none@hypv101.company.com:1

Since the push from the Hyper-V proxy adds the entries into this configuration, you may consider modifying the entries using the 'ovconfchg -edit' command.

I would suggest backing up your configuration and DB using the backup_restore.sh script as found within the product documentation before making any changes.

I hope that this is helpful.

Regards,

Mark

---
Mark Butler
Micro Focus SW Support Engineer
https://softwaresupport.softwaregrp.com/

View solution in original post

3 Replies
Highlighted
Micro Focus Expert
Micro Focus Expert

Re: Seeing duplicate hosts - can we safely delete one without affecting metrics data?

Jump to solution

Hello Aaron,

You bring up a good question.

Assuming that I understand correctly, it sounds like you have two Hyper-V hosts that are now showing twice within the list as targets even though nothing has changed within the proxy agent that was restarted.  Is this correct?

While my next example is not the same, I'm sharing it because it may make a difference.

I recently upgraded my KVM server from CentOS 7.5 to 7.7.  When I looked at my CO server today, I noticed that the KVM server was still listed as one target but the Host was listed twice when I used the pvdump command.  One entry that was current and the other that had not been updated since the date of the upgrade.  When I looked closer, I found that the SystemID entry had changed as part of the upgrade.

I bring this up simply that there may be any number of things that can cause an issue like this to occur.

Here is what I did to look more closely at the situation.

You can run the 'pvsupport.sh -grab' command to collect not only the logs but an export of the CO DB.

If you extract the resulting file on your workstation, you can go into the Support/Database/InstanceTables directory to find the inst_Infrastructure__Node.csv file.

What I normally do is open this within Excel, select the first column, use the 'Test to Columns' under the 'Data' menu, select 'Delimited' and add the "|" to the 'other' entry and then Finish.

This should make all of the entries in separate columns and then I use the column filters to isolate the entries that I'm looking for, etc.

Example:
Filter 'Role' for 'HOST'.
Filter 'HostName' for problem hosts.
Filter 'HostHostName' for problem hosts to see the guests associated with the hosts and the latest collection time.
Compare SystemID field if duplicate host entries exist.

I would look at the Guests that are part of those two problem hosts and confirm that they are showing as being updated and are not duplicated.

I know that this was not your question, but it is a way that I use to look at things more closely.

To answer your question, the 'pvconfig' command should only modify the target entries.  These entries are also located within the output of the 'ovconfget pvcd.hyperv'.

Ex.
# ovconfget pvcd.hyperv
CollectionType=REMOTE
Targets=DOMAIN\couser#hypv101v3.company.com:none@hypv101.company.com:1

Since the push from the Hyper-V proxy adds the entries into this configuration, you may consider modifying the entries using the 'ovconfchg -edit' command.

I would suggest backing up your configuration and DB using the backup_restore.sh script as found within the product documentation before making any changes.

I hope that this is helpful.

Regards,

Mark

---
Mark Butler
Micro Focus SW Support Engineer
https://softwaresupport.softwaregrp.com/

View solution in original post

Highlighted
Outstanding Contributor.
Outstanding Contributor.

Re: Seeing duplicate hosts - can we safely delete one without affecting metrics data?

Jump to solution

Mark,

To answer your first question, yes, as far as I know, nothing changed in our Hyper-V proxy configuration - it just reported a "new" name for two hosts after a restart.  In this case, the proxy is talking to SCVMM to get the list of hosts, so perhaps the hostname is being reported differently at the SCVMM level.

My main concern is that I can deleted the redundant node's collection configuration without removing the collected data from Vertica, and therefore from the CO console.  It appears from the CO console that all hostnames are forced to upper case, and are independent from the collection configuration, so this looks promising.

I will follow the steps you've provided, and see if I can validate the configuration.

Thanks,
Aaron

0 Likes
Highlighted
Outstanding Contributor.
Outstanding Contributor.

Re: Seeing duplicate hosts - can we safely delete one without affecting metrics data?

Jump to solution

Mark,

I followed your suggestion to download a configuration backup and examine the .csv file.  There were no duplicates, or entries with different collection timestamps, so it certainly appears that the collected data is intact, despite the complaints about nodes with no collections.

I removed the unnecessary collection entries using ovconfchg -edit, and the results were immediately reflected in the UI, with all hosts now showing up to date collection times.

The gap here appears to be that the appliance forces all data to uppercase hostnames when storing data, so the other components (Hyper-V proxy, agent on the appliance) should do the same, and be able to tolerate a change in case in a target's hostname.

Thanks for your help.

Aaron

0 Likes
The opinions expressed above are the personal opinions of the authors, not of Micro Focus. By using this site, you accept the Terms of Use and Rules of Participation. Certain versions of content ("Material") accessible here may contain branding from Hewlett-Packard Company (now HP Inc.) and Hewlett Packard Enterprise Company. As of September 1, 2017, the Material is now offered by Micro Focus, a separately owned and operated company. Any reference to the HP and Hewlett Packard Enterprise/HPE marks is historical in nature, and the HP and Hewlett Packard Enterprise/HPE marks are the property of their respective owners.