Wireless Access

Reply
Highlighted

Controller Process DPAgent Died

Does anybody know what the DP Agent process does on the Mobility Controllers? Received AirWave alert that process "dpagent" died and wanted to investigate further. 

 

We are running 8.5.0.5 and this is a 7210 controller operating as MD to a VMM. 

 

Currently if I look at the process table on the controller, dpagent appears to be running 99.8% CPU utilization, but don't want to restart it unless I better understand what it does and if this is normal or abnormal.

 

(MD-A) *#show processes sort-by cpu


%CPU S PID PPID VSZ RSS F NI START TIME EIP CMD
99.8 R 15699 3700 5568 2304 4 0 07:51 00:15:36 004076e0 gzip /tmp/core.4562.dpagent.A72xx_73491
68.2 S 3150 3146 12032 4800 4 0 Jan29 60-21:01:44 2ae39cc8 /mswitch/bin/syslogd -x -r -n -m 0 -f /mswitch/conf/syslog.conf
45.6 S 3920 3700 842880 362560 4 0 Jan29 40-17:01:41 2b9a9d2c /mswitch/bin/stm
12.5 S 4568 3700 42496 6080 0 0 Jan29 11-04:17:59 2b609d2c /mswitch/bin/central_agent
8.2 S 6901 6709 36480 28032 5 0 Jan29 7-09:34:40 2aed3848 nginx: worker process
6.5 S 4136 3700 522688 181504 4 0 Jan29 5-20:52:06 2b5f9d2c /mswitch/bin/fw_visibility

 

 

For comparison - here is the other controller on the same campus, same code, same model, but no alert for process dying:

 

(MD-B) *#show processes sort-by cpu


%CPU S PID PPID VSZ RSS F NI START TIME EIP CMD
98.0 R 3157 3153 12032 7488 4 0 05:24 02:45:14 2ae2ce14 /mswitch/bin/syslogd -x -r -n -m 0 -f /mswitch/conf/syslog.conf
29.5 S 3658 3428 762688 322944 4 0 05:25 00:49:30 2b9b47b4 /mswitch/bin/stm
10.8 S 4306 3428 42496 11776 0 0 05:25 00:18:14 2b609d2c /mswitch/bin/central_agent
5.3 S 3749 3428 84736 13440 4 0 05:25 00:08:59 2b569d2c /mswitch/bin/hwMon
3.7 S 3874 3428 476480 289344 4 0 05:25 00:06:15 2b5f9d2c /mswitch/bin/fw_visibility

 

It's clear that the syslog process is running at high cpu utilization, and i may try and limit it a bit for overall health, but the dpagent process is no where near the same cpu utilization on MD-B as MD-A. 

 

Thanks for the help



Michael Haring
If my answer is helpful, a Kudos is always appreciated!

Accepted Solutions
Moderator

Re: Controller Process DPAgent Died

hi Michael

dpagent is datapath agent, it's responsible for handling CLI commands like "show datapath <stuff>" where it must query the datapath to return data for CLI output.

 

in your post, dpagent is not at 99.8%, this is due to gzip compressing the coredump from a very recent crash of dpagent.

99.8 R 15699 3700 5568 2304 4 0 07:51 00:15:36 004076e0 gzip /tmp/core.4562.dpagent.A72xx_73491

 presume you have already escalated he coredump to TAC, but in the short term, you should review if you have any airrecorder scripts banging away on "show datapath" commands, and/or check the controller audit trail to see if anyone ran a show datapath related command just before the crash.

 

-jeff

View solution in original post


All Replies
Highlighted

Re: Controller Process DPAgent Died

Upon further review, it appears that on both controllers the actual dpagent process is running at 0% utilization and in sleep state:

 

(MD-A) *#show processes sort-by cpu | include dpagent

99.8 R 15699  3700   5568  2304 4   0 07:51 00:28:57 004075c8 gzip /tmp/core.4562.dpagent.A72xx_73491

 0.0 S 15698  3700  22528 11584 4   0 07:51 00:00:00 2aec9cc8 /mswitch/bin/dpagent

 

(MD-B) *#show processes sort-by cpu | include dpagent

 0.0 S  4300  3428  24192 13248 4   0 05:25 00:00:04 2aec9cc8 /mswitch/bin/dpagent

 

However, MD-A has an additional process running that references the dpagent process that is at 99.8% CPU.  Can I get clarity on what this additional process is and why it is in an R (Runnable) state? 



Michael Haring
If my answer is helpful, a Kudos is always appreciated!
Highlighted
Guru Elite

Re: Controller Process DPAgent Died

I would do a "show crashinfo" on both MDs.  After that, I would do a "tar crash" and collect the crash.tar and logs.tar with technical support and open a TAC case.

 

There is a Dpagent crash that was marked fixed in 8.5.0.5 https://www.arubanetworks.com/techdocs/ArubaOS/Consolidated_8.x_RN/Default.htm#ReleaseNotes/ResolvedIssues/resolved_8505.htm?Highlight=dpagent  Your issue might be a new and unseen manifestation that engineering has to look at.

 


*Answers and views expressed by me on this forum are my own and not necessarily the position of Aruba Networks or Hewlett Packard Enterprise.*
ArubaOS 8.5 User Guide
InstantOS 8.5 User Guide
Airheads Knowledgebase
Airheads Learning Videos
Remote Access Point Solution Guide
ArubaOS Consolidated Release Notes
ArubaOS 8 ViA VPN Solution Guide
Moderator

Re: Controller Process DPAgent Died

hi Michael

dpagent is datapath agent, it's responsible for handling CLI commands like "show datapath <stuff>" where it must query the datapath to return data for CLI output.

 

in your post, dpagent is not at 99.8%, this is due to gzip compressing the coredump from a very recent crash of dpagent.

99.8 R 15699 3700 5568 2304 4 0 07:51 00:15:36 004076e0 gzip /tmp/core.4562.dpagent.A72xx_73491

 presume you have already escalated he coredump to TAC, but in the short term, you should review if you have any airrecorder scripts banging away on "show datapath" commands, and/or check the controller audit trail to see if anyone ran a show datapath related command just before the crash.

 

-jeff

View solution in original post

Highlighted

Re: Controller Process DPAgent Died

Thank you for the explanation, I will look further into it. That information is very helpful!



Michael Haring
If my answer is helpful, a Kudos is always appreciated!
Search Airheads
cancel
Showing results for 
Search instead for 
Did you mean: