Web GUI stops responding 2.3.1



  • Hi there,

    I have just configured a 2 CARP node cluster for a datacenter deployment. During this process, especially once the XMLRPC sync was configured, the webGUI on both boxes periodically became unavailable several times. It also appears to happen when I fail one node and bring it back online, one of the boxes will loose the GUI.

    I can recover it by restarting PHP-FPM. Even when the GUI is unavailable, networking functions continue as normal so its just a crash of the front end.

    I wondered if anyone else encounters this from time to time? From memory I also saw it on a previous cluster running 2.2, occasional restarts of the web front end were required.

    I have increased the web thread count from 2 to 4 since I wondered if it was due to XMLRPC sync consuming all the web threads, but I am not sure its made a difference at all.

    Tom


  • Rebel Alliance Developer Netgate

    We see occasional reports of this but have yet to reproduce it as far as I'm aware. I have a 2.3.1 HA pair that synchronizes fine, but it is not a very complex configuration. In order for the GUI to be hung up by the synchronization, some process happening has to be crashing or causing PHP to become stuck.

    Do you have any packages installed and configured to synchronize? Or just the base?

    What features do you have enabled, in general? (e.g. IPsec, captive portal, etc)



  • Hi Jimp ,

    I have the same problem with the base package and latest version
    I enabled Captive Portal only.

    How can i debug it?


  • Rebel Alliance Developer Netgate

    You can debug it by unchecking some of the synchronization areas under System > HA Sync and see if you can determine if there is any one section specifically that, when unchecked, allows it to continue functioning indefinitely. That is assuming it's related to XMLRPC sync at all.



  • HA in uncheck mode all the time.