I am having issues with the "Update Everything" feature for the HUB

I have 12 sites registered in the Hub. When updates are available I have grown more comfortable with Update Everything.
But it seems some updates fail while others succeed. So I just keep running it until everything eventually updates. It usually takes 3-4 rounds.
Is this normal?

  • Tony G

    This issue report to WPMU DEV for the Hub was posted by Support "on my behalf" to the General Wordpress category of the support discussion forum. For all the times that I've wanted to post directly to this forum and couldn't, this isn't one of them. But since we're here...

    To complete the picture it seems when a lot of updates are queued from the Hub for a bunch of sites, the sites start to return some error 500's amongst the 200's. There are 2-3 retries but ultimately the auto process gives-up and reports failures. The current solution is simply to re-run Update Everything 3-4 times (more for more sites and more updates?) and eventually everything does update. I've asked Support to increase the count in the retry loop for each update. It would be helpful if the process re-looped on its own until there is a final, consistent number of failures for each site, which is what we would do if we kept doing it ourselves.

    So now that this is here, I hope this serves as an FYI of an issue and how it can be approached.

  • Nazmul H. Rupok

    Hi Tony

    Hope you are doing well. Sorry to hear that you are having this trouble. I have checked the chat history and your log files. I can see mostly the"Premature end of script headers: wp-load.php" and " 500 internal server errors" while trying to update the plugins at once. This usually happens when your php processes are getting killed. Seeing as how this seems to be on a shared host, the server is probably under heavy load and processes are getting killed because of resource usage.

    Since retrying the update works and all the plugins update again, so apparently your server was not able to handle all the requests at once which is not unusual. I'd suggest to perform the updates regularly or select few plugins to update at once instead of all at once. :slight_smile:

    I hope this makes sense and let me know if I could assist further from our end.

    Have a nice day. Cheers!
    Nazmul

  • Tony G

    Cross-referencing with this thread.

    I suggest the solution is not to babysit the technology but to ensire the technology accommodates a more diverse environment. Can someone tell us how many threads The Hub can open simultaneously against a single site at a given time? If there are ten plugin updates available, are ten connections launched against the site? In addition to my suggestions in the other thread for more retries, I suggest the Hub throttle at something like 4 simultaneous connections per site for updating plugins. I have no idea what a reasonable limit is or if my host or some network intermediary is throttling connections from a given client to a given host.

    But the solution is not for us to manage this process by-hand.

  • Tony G

    Status update. I've upgraded all my sites to WP4.8, PHP 5.7, activated FastCGI, and increased max_execution_time to 300s. It's still a shared host and I'm dealing with optimizations at that level as well as possible "procmonitor" termination of random processes during server load. I'm even considering taking the sites to a private server. It's probably time for me to grow out of this shared space.

    However, the best revelation today was when I checked a Content Filter plugin and found the site responding immediately afterward. The problem there is that I have 250K of filter words ... yeah, I really don't want people to be rude on my sites, in any language. :slight_smile: Remove the keywords and the site flies. Replace them and it drags. Voilà!

    So that addresses some of the instabilities. I'd still recommend that the Hub updates be a bit more resilient and informative. Handling errors is usually the last thing we add as developers, cuz as we're writing the code we believe we and the code are infallible. :slight_smile: But IMO, it's also time for the Hub code to grow beyond the developers' original (common) view of the world that everything will work as expected and that error handling is an unnecessary expense. (WINKIE!) Thanks.

  • Dimitris

    Hey there Tony,

    hope you're doing good today and don't mind chiming in here! :slight_smile:

    Is there any similarities to the websites that deal with this issue?
    I have around 10-15 testing websites, splitted in two different servers (one is a Siteground shared package) and there's been some months now I think, that I don't deal with such issues in my Hub page...
    We're keen on getting the most and best info about this, as we constantly trying to make Hub services much much better for our members!
    Your feedback here is already much appreciated! :slight_smile:

    Warm regards,
    Dimitris

  • Tony G

    Status check: Should I close this as resolved, as in "my concerns have been noted and now it's just time to let the future happen"? Or should I be waiting for some note here that "yes, we looked at the update feature and have increased the timeout values as suggested"?

    I'm in no rush. When a thread is left hanging here, I simply don't know if there is anything to do next. Actually I have noticed that more sites seem to be updating in bulk than before, though I still need to run a few times to get all of them. I just don't know where that improvement came from or if maybe my perception without any log data to support it is faulty.

    Thanks!

Thank NAME, for their help.

Let NAME know exactly why they deserved these points.

Gift a custom amount of points.