Skip to Content.
Sympa Menu

perfsonar-user - Re: [perfsonar-user] Still fighting with perfsonar-archive and maddash bundles, in both Centos7 and Ubuntu

Subject: perfSONAR User Q&A and Other Discussion

List archive

Re: [perfsonar-user] Still fighting with perfsonar-archive and maddash bundles, in both Centos7 and Ubuntu


Chronological Thread 
  • From: Andrew Lake <>
  • To: Phil Reese <>, "" <>
  • Subject: Re: [perfsonar-user] Still fighting with perfsonar-archive and maddash bundles, in both Centos7 and Ubuntu
  • Date: Mon, 26 Jun 2023 06:54:22 -0700

Hi Phil, 

Try these nightly RPMs on your MaDDash host (really just a change in the perfsonar-psconfig-maddash, but versions dependencies make the others required):

yum install https://perfsonar-dev3.grnoc.iu.edu/nightly/el/7/x86_64/perfsonar/5/packages/perfsonar-psconfig-maddash-5.0.4-0.20230626134414.el7.noarch.rpm https://perfsonar-dev3.grnoc.iu.edu/nightly/el/7/x86_64/perfsonar/5/packages/perfsonar-psconfig-maddash-devel-5.0.4-0.20230626134414.el7.noarch.rpm https://perfsonar-dev3.grnoc.iu.edu/nightly/el/7/x86_64/perfsonar/5/packages/perfsonar-psconfig-utils-5.0.4-0.20230626134414.el7.noarch.rpm


Looks like there was a bug caused by anther recent change that was making the pSConfig maddash generator have trouble figuring out which archive to use. The fix for that is in 5.0.4. I ran the RPMs above against your file and it worked. 

We might do a 5.0.4 next week. Mark is out a couple weeks so somewhat trying to hold-off unless we absolutely have to until he is back. 5.0.4 will have this fix and the fix for selinux in the archive package. 

Thanks,
Andy

On June 25, 2023 at 7:35:54 PM, Phil Reese () wrote:

Hi,

My Maddash woes continues.

I can now confirm that the Centos and Ubuntu branches produce the same br/>results, right down the  same maddash.yaml files. <

The only 'archive' solution I can get working is to build a Tooklit node br/>which has everything and use that for the archive host.&nnbsp; The bundles for br/>Centos and Ubuntu all seem to install smootthly but neither can be used br/>as a Data Source in Grafana. <

Ubuntu give  this error in Grafana:
OpenSearch error: <!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML 2.0//EN"> br/><html><head> <title>503 Serviice Unavailable</title> </head><body> br/><h1&gtt;Service Unavailable</h1> <p>The server is temporarily unable to br/>service your request due to maintenance downtime or capacityy problems. br/>Please try again later.</p> <hr> <addrress>Apache/2.4.41 (Ubuntu) Server br/>at vag-md.ufixu.com Port 4443</address> </body></html>

Centos shows this:
OpenSearch error: <!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML 2.0//EN"> br/><html><head> <title>503 Serviice Unavailable</title> </head><body> br/><h1&gtt;Service Unavailable</h1> <p>The server is temporarily unable to br/>service your request due to maintenance downtime or capacityy problems. br/>Please try again later.</p> </body></hhtml>

Further, Ubuntu is unable to start opensearch (Centos does start br/>opensearch but has the same error from the Grafana data source plugin):
● opensearch.service - OpenSearch
     Loaded: loaded (/lib/systemd/system/opensearch.service; enabled; br/>vendor preset: enabled) <
     Active: failed (Result: signal) since Sun 2023-06-25 22:38:10 UTC; br/>4s ago <
       Docs: https://opensearch.org/
    Process: 72799 br/>ExecStart=/usr/share/opensearch/bin/systemd-entrypoint -p br/>$${PID_DIR}/opensearch.pid --quiet (code=killed, sign>
   Main PID: 72799 (code=killed, signal=KILL)

Jun 25 22:38:02 vag-md systemd[1]: Starting OpenSearch...
Jun 25 22:38:10 vag-md systemd[1]: opensearch.service: Main process br/>exited, code=killed, status=9/KILL
Jun 25 22:38:10 vag-md systemd[1]: opensearch.service: Failed with br/>result 'signal'. <
Jun 25 22:38:10 vag-md systemd[1]: Failed to start OpenSearch.

A final data point is that the Maddash grid page opens, shows the br/>expected name of the dashboard but never shows any grids (the naame is br/>from the .json file).  There are no grids listed undeer the 'All Grids" br/>dropdown.  Both OS bundles show this saame behavior.

I now have a vagrant/ansible project, one for Centos and one for Ubuntu, br/>which builds 3 testpoints, an archive and md node.  This makes retesting br/>easy though it takes a long time to build aall those nodes (for some br/>reason Ubuntu takes considerable longerr!)  Happy to share if interested.

Out of curiosity, has anyone in the community been able to get the PS5 br/>v5.0.3 Archive and/or Maddash bundles working??    Would you have any br/>tips to share??

Thanks,
Phil



--
To unsubscribe from this list: https://lists.internet2.edu/sympa/signoff/perfsonar-user



Archive powered by MHonArc 2.6.24.

Top of Page