Force clearing disconnected sessions

[eliadr]

Hi.

I've a management server R81 Take 36.
I've an admin session stuck - can't login with that user or even discard, take over or publish.
Just getting "Internal error".
cpm.elg show lots of different errors around it.

Trying to remove it from mgmt_cli, results in the following error:
"Runtime error: Tried to open non existing session with id <SOME GUID>"

Any ideas? or just TAC?

[_Val_ · Accepted solution]

Please open a TAC request

[the_rock]

Im pretty sure cpstop;cpstart or reboot would fix that. However, if that was done, then I guess you may need to contact TAC.

[George_Ellis]

Zombie Thread Alert.  Same deal on 81.10 JHF 95.  Reboot did not fix it on a MDM.  Setting up to call TAC.

[the_rock]

Have you tried below command?

Andy

mgmt_cli -r true show sessions details-level full -f json | jq -r '.objects[] | select(."user-name" != "WEB_API") | .uid' | while read -r uid; do mgmt_cli -r true disconnect uid $uid; done

[George_Ellis]

We did and got 

"Runtime error: Tried to open non existing session with id <SOME GUID>"

The guid response is not in the worksession data.

mgmt_cli -r true disconnect uid UID -d DOMAINID
and
mgmt_cli -r true disconnect uid UID
 

[the_rock]

This is Provider 1?

[George_Ellis]

MDM, yes.  We will need TAC on this one.

[the_rock]

Ah, sorry, if its MDM, then not sure, hope TAC can help. I have some time later, I will set up MDM in the lab and test myself, if I make progress, will let you know either way.

Keep us posted.

Andy

[the_rock]

K, though Im working on really difficult Fortigate vpn issue, Im in parallel doing the MDS lab. I got it going, so will see how far I get. Will update you little later on the progress.

Kind regards,

Andy

[the_rock]

This is from my R81.20 jhf 26 mgmt lab

Andy

[Expert@CP-management:0]# mgmt_cli -r true show sessions details-level full -f json | jq -r '.objects[] | select(."user-name" != "WEB_API") | .uid' | while read -r uid; do mgmt_cli -r true disconnect uid $uid; done
message: "OK"

---------------------------------------------
Time: [10:05:57] 19/10/2023
---------------------------------------------
"Publish operation" succeeded (100%)
[Expert@CP-management:0]#

[George_Ellis]

It works great, but the problem is this particular UID is broken.  You get a generic "An internal error has occurred" dialog box, so went to expert cli and it fails with better information.

[the_rock]

K, just ran it on MDS level, worked 100%

See below.

Andy

[Expert@CP-Provider1:0]# uptime
11:17:49 up 14 min, 1 user, load average: 1.03, 1.24, 1.00
[Expert@CP-Provider1:0]# mgmt_cli -r true show sessions details-level full -f json | jq -r '.objects[] | select(."user-name" != "WEB_API") | .uid' | while read -r uid; do mgmt_cli -r true disconnect uid $uid; done
message: "OK"

---------------------------------------------
Time: [11:18:16] 19/10/2023
---------------------------------------------
"Publish operation" succeeded (100%)
[Expert@CP-Provider1:0]# mdsstat

CPM: Check Point Security Management Server is running and ready

+------+--------------------+-----------------+-------------+-------------+-------------+-------------+-------------+
| Type | Name | IP address | FWM | FWMHA | FWD | CPD | CPCA |
+------+--------------------+-----------------+-------------+-------------+-------------+-------------+-------------+
| MDS | - | 172.16.10.242 | up 11869 | up 11880 | up 11867 | up 7792 | up 17354 |
+------+--------------------+-----------------+-------------+-------------+-------------+-------------+-------------+
Total Domain Management Servers checked: 0 0 up 0 down
Tip: Run mdsstat -h for legend

[Expert@CP-Provider1:0]#

[Gero_Stolle]

Yes, I would also recommend open a TAC case , because your database need afixup to remove different table entries. 
There additional scripts available to do the rudimentary cleanup, but do it with TAC 🙂 

Gero