Performance Advisor stats after vol rename

[gwhite1030]

After renaming a volume from "TEST1" to "test1", performance advisor no longer gathers statistics for this volume.

Operations manager does reflect the change and all data regarding the volume is available exept performance.

When I run the command:

"dfm perf counter list filer1:volume:test1:*:*:*"

the output shows the performance counter instances for the old volume name (TEST1).

C:\>dfm perf counter list filer1:volume:test1:*:*:*
Object    Instance                 Counter                  Label1    Label2    Unit      Priv
-------   ---------                --------                 -------   -------   -----     -----
volume    TEST1                  cifs_write_data                              b_per_sec advanced
volume    TEST1                  pq_max_user_reads                            none      advanced
volume    TEST1                  nfs_other_latency                            microsec  advanced
volume    TEST1                  sys_ops                                      per_sec   basic
volume    TEST1                  pq_sys_read_limit                            none      advanced
volume    TEST1                  nfs_read_ops                                 per_sec   advanced
volume    TEST1                  throughput                                   b_per_sec basic
volume    TEST1                  pq_max_sys_reads                             none      advanced
volume    TEST1                  cifs_ops                                     per_sec   basic
volume    TEST1                  nfs_write_latency                            microsec  advanced
volume    TEST1                  pq_nvlog_limit                               none      advanced
volume    TEST1                  nfs_ops                                      per_sec   basic
volume    TEST1                  pq_nvlog_used_max                            none      advanced
volume    TEST1                  cifs_latency                                 millisec  basic
volume    TEST1                  pq_usr_weight                                none      advanced
volume    TEST1                  pq_nvlog_limit_full                          none      advanced
volume    TEST1                  nfs_latency                                  millisec  basic
volume    TEST1                  pq_sys_weight                                none      advanced
volume    TEST1                  san_ops                                      per_sec   basic
volume    TEST1                  pq_usr_sched_total                           per_sec   advanced
volume    TEST1                  other_ops                                    per_sec   basic
volume    TEST1                  cifs_write_ops                               per_sec   advanced
volume    TEST1                  nfs_read_latency                             microsec  advanced
volume    TEST1                  total_ops                                    per_sec   basic
volume    TEST1                  other_latency                                microsec  basic
volume    TEST1                  nfs_write_ops                                per_sec   advanced
volume    TEST1                  read_data                                    b_per_sec basic
volume    TEST1                  cifs_other_ops                               per_sec   advanced
volume    TEST1                  write_latency                                microsec  basic
volume    TEST1                  read_latency                                 microsec  basic
volume    TEST1                  nfs_other_ops                                per_sec   advanced
volume    TEST1                  read_ops                                     per_sec   basic
volume    TEST1                  avg_latency                                  microsec  basic
volume    TEST1                  cifs_read_data                               b_per_sec advanced
volume    TEST1                  cifs_other_latency                           microsec  advanced
volume    TEST1                  write_data                                   b_per_sec basic
volume    TEST1                  cifs_write_latency                           microsec  advanced
volume    TEST1                  nfs_read_data                                b_per_sec advanced
volume    TEST1                  cifs_read_ops                                per_sec   advanced
volume    TEST1                  san_latency                                  millisec  basic
volume    TEST1                  cifs_read_latency                            microsec  advanced
volume    TEST1                  write_ops                                    per_sec   basic
volume    TEST1                  pq_sys_sched_total                           per_sec   advanced
volume    TEST1                  pq_usr_read_limit                            none      advanced
volume    TEST1                  user_ops                                     per_sec   basic

Is there any way to update performance advisor to recognize the name change?

If necessary I am willing to purge any data related to this specific volume. Thanks.

[adaikkap]

This sounds like a bug, but did you try renaming the volume differently not just from upper case to lower case.

Regards

adai

[gwhite1030]

I have tried changing the name to something entirely different like cookie, and in this case the vol rename does appear to follow the new name and performance stats are maintained. However, on our production systems the naming convention is standardized and this is the reason for the "case" change. Does anyone know of any way to purge all the perf history for a certain object or manually tell DFM to reference the new name change as per my previous post?

[bipulm]

If you store the performance data using views, you can destroy the view using this command " dfm perf view destroy <view_name>.

You can also set the perf archive directory data to somewher else. The default location for this gloabl option "perfArchiveDir" always pointing to this

directory "/opt/NTAPdfm/perfdata". You can execute the following command for deatils:

$dfm options list | grep perf

perfArchiveDir  ----->  " /opt/NTAPdfm/perfdata"

$dfm option set perfArchiveDir = < new_location>

After that you can delete the old location. Whenever you change the location performance advisor always store the data in new location.

[adaikkap]

Hi bipul,

I think his question is, is there a way I can delete the data collected for volume TEST1 so that PA starts collecting for volume test1.

Regards

adai

[gwhite1030]

Adaikkappan, You're right. If possible I only want to remove the performance data for the affected volume (TEST1). I don't want to lose all the stats for the rest of our environment. I'm hoping that if I can remove all references to TEST1 Performance Advisor will start collecting data for test1. I just need to know how to do this... Thanks.

[harish]

Purging historical data for a given object (volume, for example) is not

supported. Also, when a volume is renamed, data collection stops for the

renamed volume which is a defect in Performance Advisor.

Regards

Harish

[gwhite1030]

Ok, how do I purge all performance history without purging ops manager historical physical stats like vol size, etc? I attempted to change the location of the performance data directory and Performance Advisor still retained the volume instance information for the old volume name. I need a way to purge the instance info.Thanks.

[harish]

Unfortunately, there is no way to purge performance history. Reducing

retention period and increasing again would be the closest alternative,

but will not help in this case since the incorrect earlier name (TEST1)

is still used by performance advisor to fetch data from the storage

system that never succeeds.

Regards

Harish