Avamar: RMCP Not Removing Checkpoints

During maintenance activities, checkpoints are not being removed. Also, if Avamar is integrated with Data Domain, snapshots are also not being expired.

admin@av-srv-prod:~/>: cplist --full
cp.20241021171415 Mon Oct 21 13:14:15 2024   valid --- del  nodes   1/1 stripes    277
cp.20241022164600 Tue Oct 22 12:46:00 2024   valid rol del  nodes   1/1 stripes    277
cp.20241022171838 Tue Oct 22 13:18:38 2024   valid --- del  nodes   1/1 stripes    277
cp.20241022193333 Tue Oct 22 15:33:33 2024   valid rol del  nodes   1/1 stripes    277
cp.20241024164621 Thu Oct 24 12:46:21 2024   valid rol ---  nodes   1/1 stripes    277
cp.20241024171054 Thu Oct 24 13:10:54 2024   valid --- ---  nodes   1/1 stripes    277
admin@av-srv-prod:~/>:

Several validated checkpoints (rolling HFS check) are displayed as "Failed" using mccli command:

admin@av-srv-prod:~/>: mccli checkpoint show --verbose
0,23000,CLI command completed successfully.
Tag               Time                    Validated Deletable Nodes Stripes Validation Start Time   Validation Finished Time Errors
----------------- ----------------------- --------- --------- ----- ------- ----------------------- ------------------------ ------
cp.20241021171415 2024-10-21 13:14:15 EDT           No        1     277     Not Validated           Not Validated            N/A
cp.20241022164600 2024-10-22 12:46:00 EDT Failed    No        1     277     2024-10-22 12:53:44 EDT 2024-10-22 13:09:46 EDT  1
cp.20241022171838 2024-10-22 13:18:38 EDT           No        1     277     Not Validated           Not Validated            N/A
cp.20241022193333 2024-10-22 15:33:33 EDT Failed    No        1     277     2024-10-22 15:42:07 EDT 2024-10-22 15:56:48 EDT  1
cp.20241024164621 2024-10-24 12:46:21 EDT Failed    No        1     277     2024-10-24 12:53:09 EDT 2024-10-24 13:08:04 EDT  1
cp.20241024171054 2024-10-24 13:10:54 EDT           No        1     277     Not Validated           Not Validated            N/A
admin@av-srv-prod:~/>: 

When the removed checkpoint (rmcp) command is performed, no checkpoints are removed.

admin@av-srv-prod:~/>: avmaint rmcp --full --ava
<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<checkpointrmlist has-approved-checkpoint="false">
  <checkpoint
    tag="cp.20241021171415"
    deleted="false"
    ddr-deleted="false"/>
  <checkpoint
    tag="cp.20241022164600"
    deleted="false"
    ddr-deleted="false"/>
  <checkpoint
    tag="cp.20241022171838"
    deleted="false"
    ddr-deleted="false"/>
  <checkpoint
    tag="cp.20241022193333"
    deleted="false"
    ddr-deleted="false"/>
  <checkpoint
    tag="cp.20241024164621"
    deleted="false"
    ddr-deleted="false"/>
  <checkpoint
    tag="cp.20241024171054"
    deleted="false"
    ddr-deleted="false"/>
</checkpointrmlist>

On Data Domain, snapshots do not expire automatically. They must be expired manually:

avboost@dd-srv-prod# snapshot list mtree /data/col1/avamar-1234567890
Snapshot Information for MTree: /data/col1/avamar-1234567890
----------------------------------------------
Name                Pre-Comp (GiB)   Create Date         Retain Until        Status
-----------------   --------------   -----------------   -----------------   -------
cp.20241015171741          69287.4   Oct 15 2024 13:19   Oct 22 2024 13:13   expired
cp.20241015194118          69287.4   Oct 15 2024 15:43   Oct 22 2024 13:13   expired
...
...
cp.20241020164654          65247.4   Oct 20 2024 12:49
cp.20241020171602          65262.9   Oct 20 2024 13:18
cp.20241021164757          65257.4   Oct 21 2024 12:50
cp.20241021171415          65272.9   Oct 21 2024 13:16
cp.20241022164600          65280.0   Oct 22 2024 12:48
-----------------   --------------   -----------------   -----------------   -------
...
avboost@dd-srv-prod# 

Another observed behavior is slowness when running commands on the Avamar server. Despite the server not running any tasks or backups, the load average remains high.

Several factors can cause this behavior. All issues were identified after thoroughly analyzing the processes (using top or ps -ef) running on the Avamar server. Some scenarios include:

• Old Perl processes
• Outdated custom replication
• Custom reports
• Old Avtar processes

Evidences are found in some scenarios:

admin    15007  0.0  0.0   9664  2812 ?        Ss    2023   0:00 bash -c export TERM=${TERM:-dumb} ; /usr/bin/ssh-agent /tmp/dpnctl-run-self.14963.aux
admin    15042  0.0  0.0   9528  2192 ?        S     2023   0:00  \_ /bin/bash /tmp/dpnctl-run-self.14963.aux
admin    15043  0.0  0.0  30792   680 ?        Ss    2023   0:52      \_ /usr/bin/ssh-agent /tmp/dpnctl-run-self.14963.aux
admin    15049 99.6  0.1  81996 39340 ?        R     2023 272656:21      \_ /usr/bin/perl /usr/local/avamar/bin/dpnctl --rerun --mcs_user=root stop 

admin    26975     1  0  80   0 -  3440 -      Oct08 ?        00:00:00 bash -c ./avReplication.40 --report --csv --quiet
admin    27290 25935  0  80   0 -  3440 -      Oct08 ?        03:55:24 bash -c ./avReplication.40 --quiet --report --short-status
admin    27761 26975  0  80   0 -  3440 -      Oct08 ?        03:50:39 bash -c ./avReplication.40 --report --csv --quiet

root      9046  0.0  0.0 314212  6792 ?        SNl  Nov08   0:00 /usr/local/avamar/bin/avtar.bin --vardir=/usr/local/avamar/var --bindir=/usr/local/avamar/bin --sysdir=/usr/local/avamar/etc --sysdir="/usr/l
root     20385  0.0  0.0 314212  6624 ?        SNl  Nov08   0:00 /usr/local/avamar/bin/avtar.bin --vardir=/usr/local/avamar/var --bindir=/usr/local/avamar/bin --sysdir=/usr/local/avamar/etc --sysdir="/usr/l
root     22784  0.0  0.0 314212  6544 ?        SNl  Nov08   0:00 /usr/local/avamar/bin/avtar.bin --vardir=/usr/local/avamar/var --bindir=/usr/local/avamar/bin --sysdir=/usr/local/avamar/etc --sysdir="/usr/l

1. Log in to the Avamar server as admin switch to the root:

su -

2. Run the following commands to thoroughly analyze the processes:

top
ps -aux --forest
ps -ef

3. Once the process is identified, terminate it using the process ID (PID):

kill <pid>

4. If the process does not terminate, force it:

kill -9 <pid>

5. The commands should start working faster again.

6. Run RMCP:

avmaint rmcp --full --ava

7. The following two commands show the checkpoint correctly again:

cplist --full
mccli checkpoint show --verbose

Example:

admin@av-srv-prod:~/>: cplist --full
cp.20241024164621 Thu Oct 24 12:46:21 2024   valid rol ---  nodes   1/1 stripes    277
cp.20241024171054 Thu Oct 24 13:10:54 2024   valid --- ---  nodes   1/1 stripes    277
admin@av-srv-prod:~/>: 

admin@av-srv-prod:~/>: mccli checkpoint show --verbose
0,23000,CLI command completed successfully.
Tag               Time                    Validated Deletable Nodes Stripes Validation Start Time   Validation Finished Time Errors
----------------- ----------------------- --------- --------- ----- ------- ----------------------- ------------------------ ------
cp.20241024164621 2024-10-24 12:46:21 EDT Validated No        1     277     2024-10-24 12:53:09 EDT 2024-10-24 13:08:04 EDT  0
cp.20241024171054 2024-10-24 13:10:54 EDT           No        1     277     Not Validated           Not Validated            N/A
admin@av-srv-prod:~/>: 

8. Ensure that snapshots on Data Domain show "expired" Status.