Ad Widget

Collapse

Strange map label problem in 5.0.4

Collapse
X
 
  • Time
  • Show
Clear All
new posts
  • mike.patchen
    Junior Member
    • Apr 2014
    • 7

    #1

    Strange map label problem in 5.0.4

    I've got three Zabbix servers (Dev, Test, and Prod) with only one of them having this problem. The problem is in maps (any map) all the icon labels are in the upper left corner of the map, on top of on another. I've tried dumping the Prod database and loading it into Test and the behavior does not follow - i.e. the same map in Test looks normal with labels under the icon as they should be.

    Here is an example of the issue:
    Click image for larger version

Name:	Bad Map.jpg
Views:	232
Size:	7.0 KB
ID:	410753

    And here is the exact same map from Test:
    Click image for larger version

Name:	Good map.jpg
Views:	214
Size:	8.3 KB
ID:	410754

    Anyone have any ideas on what may be causing this? It wouldn't be so big of a deal but, of course, it has to be happening on my Prod server.
  • mike.patchen
    Junior Member
    • Apr 2014
    • 7

    #2
    Ok, plot thickens. I was able to get 10 minutes downtime and swung my Dev front end to the Prod database. Maps display correctly in this configuration.

    I've verified /etc/httpd/conf.d/zabbix.conf, /etc/zabbix/web/zabbix.conf.php and /etc/php-fpm.d/zabbix.conf are all identical between both systems.

    Both systems patched to CentOS 8.1.1911. Due to restrictions on our Corporate proxies I have to maintain my own repo server so I know all systems have exactly the same packages.

    Anything else anyone could think of? I'm about to just say to heck with it and build a new server and swing the IP address over - see if that fixes it. I'd really rather not go that extreme as this is a production system.

    Comment

    • tim.mooney
      Senior Member
      • Dec 2012
      • 1427

      #3
      Originally posted by mike.patchen
      Due to restrictions on our Corporate proxies I have to maintain my own repo server so I know all systems have exactly the same packages.
      Eh, that just means they all could have the same set of packages installed; it doesn't guarantee that prod has the exact same set as the other systems.

      In your situation, I would start by generating a sorted list of installed packages on prod and again on dev, and then diff them to see if prod is either missing something or has something extra.

      If that's not the problem, report back.

      Comment

      • mike.patchen
        Junior Member
        • Apr 2014
        • 7

        #4
        Excellent point, I would change your couldto should though. But you are correct, there are a handful of packages that differ in minor revs (i.e. 647.2.11-2 vs 647.2.24-1). I'll update tonight during maintenance window and see if that solves the problem.

        Comment

        • mike.patchen
          Junior Member
          • Apr 2014
          • 7

          #5
          Sadly that did not fix the problem. I've confirmed all packages are the same between the systems. Any other ideas out there?

          Comment

          • mike.patchen
            Junior Member
            • Apr 2014
            • 7

            #6
            Solved!

            Issue was with Browser cache. I had ruled this out initially because I tried from different browsers (AND different computers, and different logins) but something must have been wrong with my profile at some point that caused it to not work correctly. I finally realized it when on my monthly enterprise-wise Zabbix Q&A session this morning and apologizing for the map issues since the migration to version 5. A few people agreed it was a problem, but several others asked me "what map issues?".

            In case you're wondering, the migration was an export/import and recreate some stuff (well, a lot of stuff) not a DB upgrade. We had been running V3.2 and it wasn't really built like it should have been for a company of our size so when installing V5 we decided to start over fresh. Lots of API calls in custom scripts to migrate things over, but the migration only took a few hours as opposed to the amount of time it would have taken to upgrade a 2TB database. Yes, we lost history and trends but decided it was worth it to rebuild it for the future.

            Comment

            Working...