perfSONAR User Q&A and Other Discussion

["John Tai (johntai)" <>]

Thanks Andrew!  The reinstall fixed the issue.

 

The failure scenario described exactly why my laptop install worked cleanly since it’s VM is dhcp based, while our lab setup requires static IPs which are manually added in after the perfsonar VM was booted up.

 

Regards,

John

 

============

John Tai, Jr.

408-853-9540

 

From: Andrew Lake [mailto:]
Sent: Friday, April 21, 2017 5:28 AM
To: John Tai (johntai) <>;
Subject: Re: [perfsonar-user] perfsonar 4.0 pscheduler archiver does not start behind firewall w/ "pscheduler" does not exist FATAL Error

 

Hi John,

 

Try running “yum reinstall pscheduler-server”. A couple others have reported a similar issue and that fixed it. Because of some limitations of what the packages are allowed to do during the ISO install phase there are a few setup things we have to do on first boot. We think what’s happening is they are not completing all the way if networking is not all the way up yet. We are working on building and testing a new ISO that fixes this, but running “yum reinstall pscheduler-server” should get you where you need to be.

 

Thanks,

Andy

 

 

 

On April 21, 2017 at 12:59:19 AM, John Tai (johntai) () wrote:

Hi Experts!

 

I’m a perfsonar newbie and installed the pS-Toolkit-4.0-CentOS7-FullInstall-x86_64-2017Apr17.iso on an ESXi VM behind a lab firewall and cannot get the pscheduler archiver to run.

 

/var/log/pscheduler/pscheduler.log constantly restarts with the below error “OperationalError: FATAL:  role "pscheduler" does not exist”.  More detail log below.

 

The follow updates to the CentOS7 were done to handle firewall issues:

 

1)      /etc/profile.d/proxy.sh                 !!! set http_proxy/https_proxy to allow “only” http/https traffic on ports 80/443 to go thru firewall

2)      /etc/resolv.conf                                               !!! updated for DNS servers

3)      /etc/sysconfig/network-scripts/ifcfg-eth0  !!! updated for static IP, Gateway

4)      /etc/ntp.conf                                    !!! updated to use a lab NTP server (ntp is correctly working)

 

 

Running the same ISO using a laptop VMware Player VM on a corporate network that does not have a strict firewall policy initially had similar failures, but manually running “systemctl start pscheduler-archiver” restarted the archiver successfully and everything started to work. When a perfsonar VM runs in our lab network behind the stricter FW, the above error always occurs and archiver never starts, even after issue the “systemctl restart  pscheduler-archiver” several times

 

Any help is appreciated!

 

 

Here’s the detailed pscheduler.log with “pscheduler debug on”.

I did follow the Trouble Shooting link without any luck : http://docs.perfsonar.net/troubleshooting_overview.html

 

Apr 20 21:41:39 Perfsonar-Node journal: safe_run/scheduler ERROR    Waiting 30.75 seconds before restarting

Apr 20 21:41:39 Perfsonar-Node journal: safe_run/runner ERROR    Restarting

Apr 20 21:41:39 Perfsonar-Node journal: safe_run/ticker ERROR    Restarting

Apr 20 21:41:39 Perfsonar-Node journal: ticker DEBUG    Debug started

Apr 20 21:41:39 Perfsonar-Node journal: ticker DEBUG    Debug started

Apr 20 21:41:39 Perfsonar-Node journal: ticker DEBUG    QM: Started

Apr 20 21:41:39 Perfsonar-Node journal: ticker DEBUG    QM: Started

Apr 20 21:41:39 Perfsonar-Node journal: safe_run/runner ERROR    Program threw an exception after 0:00:00.007574

Apr 20 21:41:39 Perfsonar-Node journal: safe_run/runner ERROR    Exception: OperationalError: FATAL:  role "pscheduler" does not exist#012#012Traceback (most recent call last):#012  File "/usr/lib/python2.7/site-packages/pscheduler/saferun.py", line 41, in safe_run#012    function()#012  File "/usr/libexec/pscheduler/daemons/runner", line 909, in <lambda>#012    pscheduler.safe_run(lambda: main_program())#012  File "/usr/libexec/pscheduler/daemons/runner", line 719, in main_program#012    db = pscheduler.pg_connection(dsn)#012  File "/usr/lib/python2.7/site-packages/pscheduler/db.py", line 35, in pg_connection#012    pg = psycopg2.connect(dsn)#012  File "/usr/lib64/python2.7/site-packages/psycopg2/__init__.py", line 164, in connect#012    conn = _connect(dsn, connection_factory=connection_factory, async=async)#012OperationalError: FATAL:  role "pscheduler" does not exist

Apr 20 21:41:39 Perfsonar-Node journal: safe_run/runner ERROR    Waiting 30.75 seconds before restarting

Apr 20 21:41:39 Perfsonar-Node journal: safe_run/ticker ERROR    Program threw an exception after 0:00:00.011558

 

Regards,

John

 

============

John Tai, Jr.

408-853-9540