Problems with REST API and asynchronus ucs creation

Hi all,

I'm struggling using asynchronus tasks to generate a UCS. I've been through the API reference, and tried exactly what they suggest but something isn't right. I've dug through logs, watched lsof and top while executing the commands to see what happens, and I can't figure out why this isn't working.  Searching devcentral hasn't been useful for me either.

I've tried this with 4 different devices, and all exhibit the same behaviour.

I send the POST to start the task, and I get confirmation of the task number and it's state.

curl -k -u apiadmin:<redacted> -H "Content-Type: \
application/json" -X POST -d \
'{
"command": "save",
"name": "myUcs"
}' \
https://192.168.240.15/mgmt/tm/task/sys/ucs

{"_taskId":1692187396506186,"_taskState":"STARTED","_taskTimeInStateMs":0,"_taskResultLink":"https://localhost/mgmt/tm/task/sys/ucs/1692187396506186/result?ver=16.1.3.3","selfLink":"https://localhost/mgmt/tm/task/sys/ucs/1692187396506186?ver=16.1.3.3"}%

_____________

Just to check on the task, I get the status and it shows as STARTED which we would expect. It makes no difference in the end result if I check the task before putting into VALIDATING or not.

❯ curl -k -X GET -u apiadmin:<redacted> https://192.168.240.15/mgmt/tm/task/sys/ucs/1692187396506186

{"selfLink":"https://localhost/mgmt/tm/task/sys/ucs/1692187396506186?ver=16.1.3.3","_taskId":1692187396506186,"_taskState":"STARTED","_taskTimeInStateMs":28000,"_taskResultLink":"https://localhost/mgmt/tm/task/sys/ucs/1692187396506186/result?ver=16.1.3.3"}%

_____________

I PUT the taskstate to VALIDATING, and I get confirmation that the task will start.

curl -k -u apiadmin:<redacted> -H "Content-Type: \
application/json" -X PUT -d \
'{
"_taskState":"VALIDATING"
}' \
https://192.168.240.15/mgmt/tm/task/sys/ucs/1692187396506186
{"code":202,"message":"Task will execute asynchronously.","errorStack":[]}%

_____________

Once that happens, the API seems to hang a bit as I can't make new API calls for 30 seconds to a minute, and during that time both java and gzip are churning away on a file in /var/tmp/imxxxxx/configsync-2.0-1-Linux-3.10.0862.14.4.7..8664.im. In this last case, the directory was im22934, which doesn't correspond to my requested task that I can tell.

im22934:
total 438M
drwxr-xr-x. 2 root root 4.0K Aug 16 07:06 .
drwxrwxrwt. 31 root root 12K Aug 16 07:13 ..
-rw-r--r--. 1 root root 438M Aug 16 07:05 configsync-2.0-1-Linux-3.10.0862.14.4.7..8664.im

From support, these are orphaned attempts at creating a UCS so the process is starting, but dieing and I don't know why : https://my.f5.com/manage/s/article/K49300144

curl -k -X GET -u apiadmin:<redacted> https://192.168.240.15/mgmt/tm/task/sys/ucs/1692187396506186

{"code":404,"message":"Task not found - ID: 1692187396506186 user: apiadmin","errorStack":[],"apiError":1}%

_____________

I've been through logs and not found any errors. I've tried different users, with and without underscores in the name. Nothing seems to want to work for me.

What am I doing wrong?