perfSONAR User Q&A and Other Discussion

[Michael Petry <>]

I don't seem to be able to backup the database after an upgrade to 4.0-0.7.rc1.el6

/usr/lib/perfsonar/scripts/ps-toolkit-migrate-backup.sh -d ~petry/foo_data_backup

Backing-up users...[SUCCESS]

Backing-up groups...[SUCCESS]

Backing-up passwords...[SUCCESS]

Backing up administrative users...[SUCCESS]

Backing-up toolkit configuration...[SUCCESS]

Backing-up bwctl-server configuration...[SUCCESS]

Backing-up owamp-server configuration...[SUCCESS]

Backing-up NTP configuration...[SUCCESS]

Backing-up cassandra data...Unable to snapshot cassandra database

Checking the script, it seems that the call to nodetool is failing

nodetool snapshot esmond -t esmond_snapshot

Failed to connect to '127.0.0.1:7199': Connection refused

The cassandra.log doesn't seem to indicate anything wrong from first glance.

INFO 15:57:09,753 Enqueuing flush of Memtable-local@628785744(84/840 serialized/live bytes, 4 ops)

 INFO 15:57:09,754 Writing Memtable-local@628785744(84/840 serialized/live bytes, 4 ops)

 INFO 15:57:09,768 Completed flushing /var/lib/cassandra/data/system/local/system-local-jb-84-Data.db (116 bytes) for commitlog position ReplayPosition(segmentId=1475783818984, position=191038)

 INFO 15:57:09,772 Compacting [SSTableReader(path='/var/lib/cassandra/data/system/local/system-local-jb-84-Data.db'), SSTableReader(path='/var/lib/cassandra/data/system/local/system-local-jb-81-Data.db'), SSTableReader(path='/var/lib/cassandra/data/system/local/system-local-jb-83-Data.db'), SSTableReader(path='/var/lib/cassandra/data/system/local/system-local-jb-82-Data.db')]

 INFO 15:57:09,807 Enqueuing flush of Memtable-local@2014399804(10064/100640 serialized/live bytes, 259 ops)

 INFO 15:57:09,809 Writing Memtable-local@2014399804(10064/100640 serialized/live bytes, 259 ops)

 INFO 15:57:09,849 Compacted 4 sstables to [/var/lib/cassandra/data/system/local/system-local-jb-85,].  6,261 bytes to 5,717 (~91% of original) in 46ms = 0.118525MB/s.  4 total partitions merged to 1.  Partition merge counts were {4:1, }

 INFO 15:57:09,853 Completed flushing /var/lib/cassandra/data/system/local/system-local-jb-86-Data.db (5301 bytes) for commitlog position ReplayPosition(segmentId=1475783818984, position=203003)

 INFO 15:57:09,911 Node localhost/127.0.0.1 state jump to normal

esmond.log seems that same

2016-10-06 15:57:43,826 [INFO] /usr/lib/esmond/esmond/cassandra.py: Connected to ['localhost:9160']

2016-10-06 15:57:45,548 [INFO] /usr/lib/esmond/esmond/cassandra.py: Checking/creating column families

2016-10-06 15:57:45,550 [INFO] /usr/lib/esmond/esmond/cassandra.py: Schema check done

2016-10-06 15:57:45,560 [INFO] /usr/lib/esmond/esmond/cassandra.py: Connected to ['localhost:9160']

Any ideas?

Mike

On Thu, Oct 6, 2016 at 12:11 PM, John Simpkins <> wrote:

Adding to the thread, I'm finding that attempts to patch the CentOS 6 box that upgraded from 3.5 to 4.0RC1 are met with the following error:

psql: FATAL:  no pg_hba.conf entry for host "[local]", user "postgres", database "postgres", SSL off

warning: %post(pscheduler-test-trace-1.0-0.8.rc1.el6.noarch) scriptlet failed, exit status 2

Anyone else encountered this?

On Wed, Oct 5, 2016 at 4:34 PM Bill Owens <> wrote:

On Tue, Oct 04, 2016 at 09:13:15AM -0700, Andrew Lake wrote:
>    Hi Bill,
>    On CentOS 7 we no longer use system accounts in the psadmin group with
>    mod_auth_shadow in apache to create web user accounts. Instead it uses a
>    htpasswd file and doesn’t link it to the system accounts at all. I think
>    nptoolkit-configure.py should properly add the web user to the htpasswd
>    file, but is it giving you an error? or does it act like it completes but
>    then you can’t login?

No, it threw this error:

[owens@bss532 ~]$ sudo /usr/lib/perfsonar/scripts/nptoolkit-configure.py

perfSONAR Toolkit customization script

1. Change Timezone
2. Manage Users
0. exit

Make a selection: 2

Welcome to the perfSONAR Toolkit user administration program.
This program will help you administer users.
You may configure any of the options below with this program:
    1. Add a new user
    2. Delete a user
    3. Change a user's password
    0. exit
Make a selection: 1

Enter the user whose account you'd like to add. Just hit enter to exit: psweb
Should this user be able to login via SSH? [yes] no
Should this user be able to login to the web interface? [yes] yes
adduser: group 'psadmin' does not exist
Error: couldn't add user psweb

You may configure any of the options below with this program:
    1. Add a new user
    2. Delete a user
    3. Change a user's password
    0. exit


I think that's really just a documentation problem, though; the instructions need to be changed to tell you to run the new script instead.

Now I'm fighting with the machine not wanting to run any regular bandwidth tests; I can get bwctl to work manually, and I can test against it with my 3.5 box, but it won't initiate anything. Oh, and after the most recent reboot it won't allow me to log into the web interface anymore. So I'm going home for the day ;)

Bill.