SnapCreator create directory /tmp/pdk-root-<process_id>

[OLIVIERSTUDER]

Hi,

When a job is started, a directory pdk-root-<process_id> is create in /tmp. And this directory was not removed at the end of the job.

Some body can help me ?

Regards

/Olivier

[ktenzer]

In Snap Creator 4.0 the CLI and Agent created this directory in /tmp. These are the libraries and are things that those binaries use. The directory is not removed as it is reused, so unless you delete it or it gets deleted by /tmp.

In Snap Creator 4.1 only the CLI will create this directory in /tmp as the Agent is now Java which does things differently.

What is the problem with this dir not being removed?

Keith

[OLIVIERSTUDER]

Hi,

Yes, I can remove this directory with rm command. But each time the job run, a new directory was create. So this direcory need to be delete manualy. It's not automaticly.

I give you the contains of this directory:

agetestdb1:/tmp/pdk-root-7141 # pwd

/tmp/pdk-root-7141

agetestdb1:/tmp/pdk-root-7141 # ls -l

total 2532

-r-xr-xr-x 1 root root  171015 Jan 29 13:58 VIM25Runtime.pm

-r-xr-xr-x 1 root root 1566028 Jan 29 13:58 VIM25Stub.pm

-r-xr-xr-x 1 root root   87988 Jan 29 13:58 VIM2Runtime.pm

-r-xr-xr-x 1 root root  741860 Jan 29 13:58 VIM2Stub.pm

agetestdb1:/tmp/pdk-root-7141 #

I use the web application to manage the scheduled job.

I use SnapCreator 4.1.0c.

/Olivier

[ktenzer]

Ah yes the VMware plugin and other perl plugins unpack things to /tmp. As I said they arent deleted but rather re-used. If they arent there then they are unpacked.

There is no point in deleting them. In addition we are talking a few MBs. I still dont understand the concern? As I said every time a backup runs these libraries are used. We could unpack them under somewhere else but we chose /tmp since they are important only at runtime. This is not something we can change or will change in 4.1 but we can consider this in future. Again I would like to understand the problem this creates?

Keith

[OLIVIERSTUDER]

Hi,

My problem is that when a job is started, they are a new /tmp/pdk-root-<new_process_id> directory created.

A list of the contains of /tmp:

drwxr-xr-x  2 root	root  4.0K Jan 29 12:25 pdk-root-730
drwxr-xr-x  2 root	root  4.0K Jan 29 12:26 pdk-root-963
drwxr-xr-x  2 root	root  4.0K Jan 29 12:27 pdk-root-1053
drwxr-xr-x  2 root	root  4.0K Jan 29 12:27 pdk-root-1296
drwxr-xr-x  2 root	root  4.0K Jan 29 12:28 pdk-root-1386
drwxr-xr-x  2 root	root  4.0K Jan 29 12:28 pdk-root-1407
drwxr-xr-x  2 root	root  4.0K Jan 29 12:29 pdk-root-1477
drwxr-xr-x  2 root	root  4.0K Jan 29 12:29 pdk-root-1495
drwxr-xr-x  2 root	root  4.0K Jan 29 13:23 pdk-root-5224
drwxr-xr-x  2 root	root  4.0K Jan 29 13:23 pdk-root-5243
drwxr-xr-x  2 root	root  4.0K Jan 29 13:55 pdk-root-6881
drwxr-xr-x  2 root	root  4.0K Jan 29 13:56 pdk-root-6952
drwxr-xr-x  2 root	root  4.0K Jan 29 13:58 pdk-root-7129
drwxr-xr-x  2 root	root  4.0K Jan 29 13:58 pdk-root-7141
drwxr-xr-x  2 root	root  4.0K Jan 29 14:23 pdk-root-8791
drwxr-xr-x  2 root	root  4.0K Jan 29 14:23 pdk-root-8805
drwxr-xr-x  2 root	root  4.0K Jan 29 14:24 pdk-root-9255
drwxr-xr-x  2 root	root  4.0K Jan 29 14:24 pdk-root-9282
drwxr-xr-x  2 root	root  4.0K Jan 29 14:31 pdk-root-9885
drwxr-xr-x  2 root	root  4.0K Jan 29 14:32 pdk-root-10126
drwxr-xr-x  2 root	root  4.0K Jan 29 14:32 pdk-root-10192
drwxr-xr-x  2 root	root  4.0K Jan 29 14:33 pdk-root-10431
drwxrwxrwt  5 root	root  4.0K Jan 29 14:36 .com_ibm_tools_attach
drwxr-xr-x  2 root	root  4.0K Jan 29 14:45 pdk-root-11341
drwxr-xr-x 25 root	root  4.0K Jan 29 14:45 pdk-root

Regards

/Olivier

[ktenzer]

Ok I see the issue. Probably we can make some improvements to our process handling. We will look at doing this in SC 4.2 and I will open a defect.

I would open an NGS case if you need this addressed sooner.

As a workaround till we have fix you can execute an OS command through snap creator to delete the directories or do it manually through OS.

POST_NTAP_CMD01=rm -rf /tmp/pdk*

Regards,

Keith

[ktenzer]

Disregard what I said about POST_NTAP_CMD that isnt a good idea if you have multiple jobs running then they could step on each other and cause backup failures since the data in /tmp is used at runtime.

Regards,

Keith

[OLIVIERSTUDER]

Hi,

Actually this is a test server. So I can able to delete this directroy manually.

It's also possible to create a cron job.

Regards

/Olivier

[ktenzer]

A cron job is a decent idea ideally one that isnt running when a backup is scheduled. I am raising this issue and pushing for this fix to go into a SC 4.1 patch otherwise we will fix this in SC 4.2.

If this is urgent fix for you NGS is your best route as they can escalate directly to engineering.

Thanks for bringing this to our attention

Keith

[ktenzer]

Good news we are going to try and address this issue in SC 4.1 patch release due out in a few weeks.

Keith