Cap'n Refsmmat

Administrators
  • Content count

    11781
  • Joined

  • Last visited

Community Reputation

1343 Glorious Leader

About Cap'n Refsmmat

  • Rank
    Mr. Wizard
  • Birthday 11/26/1991

Contact Methods

  • Website URL
    https://www.refsmmat.com

Profile Information

  • Gender
    Male
  • Location
    Texas
  • College Major/Degree
    BSc Physics, the University of Texas; currently PhD Statistics, CMU
  • Favorite Area of Science
    Statistics and physics.
  • Occupation
    Studying statistics
  1. In the United States, a lot of police and fire radio systems operate in the 800 MHz band, which has sections reserved for public safety use. https://www.fcc.gov/general/800-mhz-spectrum
  2. Current events

    What types of topics or discussions do you think would belong in such a section that wouldn't be covered by the Politics section?
  3. Infection

    Could you be specific about the error message and any details it provides? Oddly enough, Googling "Googleheads" doesn't turn up any information that would help me track this down.
  4. Referring to posts in threads

    If you click the "Share this post" button at the top right corner of a post (the three dots linked by two lines), you'll get a direct link to the post, which you could use to refer to a specific post.
  5. IPB is Invision Power Board, the old name for the software we use. (Eventually they rebranded as Invision Community or something.) There are two separate problems we're experiencing: The new version of IPB was saving a bunch of cache data to the hard disk with every request, causing enormous load on the hard drives. That seems to be part of why SFN was so slow -- every request was delayed waiting for its turn to use the hard disk. Dave switched it to use the database instead, which can be much smarter about keeping the data in memory instead of on disk. The "bad gateway" errors are a bug in PHP which make it crash occasionally. Our HTTP server, nginx, notices that PHP has crashed and failed to produce a webpage, and reports "bad gateway" to you. If Dave's tweak is right, it should fix problem #1 but not problem #2. I've searched for solutions to problem #2 and haven't found anything yet -- either we're the only people experiencing the problem or it's an artifact of something unique in our setup. But if you encounter the "bad gateway" errors you should be able to refresh and get the page loaded quickly. We'll monitor the changes over the next couple days and see if problem #1 is conclusively solved or if we need to make more tweaks. Big thanks to Dave for spotting the issue and fixing it.
  6. Was it this page? http://www.scienceforums.net/online/ You get that by hitting "See full list" on the Who's Online box.
  7. Your Heart's Desire

    Their excuse is that post counts can vary. Moderators can see posts that were hidden for breaking the rules; posts may be deleted after members already referred to them by number; and posts may be in the mod queue awaiting approval. I could refer to post #48, but its number may be different by the time you look at it, because post #47 was deleted. I suppose every post could be assigned a permanent number, displayed at the top right, and if #47 is deleted the thread simply skips from 46 to 48, but then you could see when we delete things, and the more censorious forums would get upset that their members could see behind the curtain.
  8. Your Heart's Desire

    The makers of our forum software believe they know better than we do, and don't intend to create an option for adding back post numbers. This is one of the many reasons why I hate all forum software. However, if you click the little Share icon at the top right of a post, you can get a direct link to the post, which you can use to send people directly to the post you're referring to.
  9. The advertisements (Google AdSense) should be asynchronous: waiting for the ads to load should never block the rest of the page from loading. I believe the ads do make periodic requests while a page is open, though, so if you're waiting for another page to load you may see some ad requests appearing in the mean time.
  10. Yeah, I was trying to debug things further and screwed up for a while. Sorry about that. Should be running fine now, though I'm still puzzled by the high latency.
  11. Just made another tweak: allowed nginx (the web server software) to use 16 processes instead of just 4, since we do have 16 cores available on the server. Perhaps that will decrease some of the latency we've been seeing, where it takes multiple seconds before the server even replies to a request.
  12. I just killed some idle processes on the server and upped the PHP opcode cache memory limit to 128MB instead of 64MB (it was full). Both should help a bit. I'll check in again tomorrow to see if the memory limit is adequate and see if there's anything more I can do.
  13. Yeah, there's a PHP bug that makes it crash occasionally, and you get a 502 when that happens. It seems to happen every few minutes, and one random person trying to load a page when it happens is the lucky winner. I wasn't able to find any bug reports that matched the crash, but I did just upgrade our PHP install, and will watch to see if that fixes it. Backups just ran and sucked up a lot of CPU. Also I was running the upgrades. Let's see how the next day or so goes and that'll tell us if it's fixed.
  14. It looks like we were getting bursts of traffic faster than we could start PHP processes, and didn't have enough spare processes. I bumped the maximum number of spare processes up to 25 (from 10) and the overall cap up to 50. I'm not sure if that will solve the entire problem, but it should prevent some of the Bad Gateway errors. Let me know what you experience over the next day or so.
  15. I'll look into this soon. We had problems soon after the upgrade, which we were able to fix, but I'm not sure what could be making the site slow now.