Jump to content
Banner by ~ Ice Princess Silky

DDR

Administrator
  • Posts

    52
  • Joined

  • Last visited

Everything posted by DDR

  1. Hm, I'll have to look into it. I'm working on an email link issue right now, I'll see about fixing signatures next.
  2. Banned for not eating pizza of your own free will.
  3. Banned for excessively derogatory manner.
  4. 2018-01-31: Removed the e-commerce integrations from the VM - apparently when I copied over the database, I got all the posts but also the the module that bills people… which was set to bill people. Now that the initial shenanigans are over, things seem to actually be quite quiet on the sysadmin front here. Updates come down the pipe from Invision occasionally, and I upload them to MLPF. I'm starting a new job tomorrow, so hopefully I'll have more time to poke around and fix some of the long-term issues we've been having. My predecessor(s) left me very bad notes, and I'm worried that we've lost a lot of experience of how to modify and fix the forum software. In addition, my dev VM doesn't actually work very well right now. (I guessed, badly, how to get the forum up and running when I set it up.) So while I can view the home page with the dev vm, I can't actually click through to any sub-forums or posts. Only the admin panel seems to work—which is enough to test the updates—but not enough if I'm going to reproduce and fix our standing issues. I guess the next things I'm going to work on are: - figure out what's broken - file them as bugs in the internal bug-tracker - properly remake the dev vm from the ground up, now that I know what I'm doing I have made some good progress documenting mlpf in the internal wiki, so the next person to come along won't be starting from scratch. There's still a lot more to go, though. I swear, 80-90% of software is just talking to people or writing things for people to read.
  5. Banned for not colouring Canada green in the quoted image.
  6. I agree. That behaviour definitely seems… suboptimal, to say the least.
  7. Thanks! Good to hear it's all working well. @Hieroklicious That's a good muffin. Kinda buggy though!
  8. 2018-01-11: Did a routine upgrade of the forums. Everything went quickly and smoothly, without issue, both in the VM and here in production. "So this is what it's supposed to be like…" I also added cron job to the VM, so it should actually do its background processing tasks now. Set the default editor to Vim.
  9. Hi everyone, I've upgraded the forums to the latest version. It should be much smoother this time, but as always let us know if there's any issues! Thanks!
  10. Maybe add attachment by clicking the "choose files…" link at the bottom of the editor might work. Yeah; same here. @Feld0 is working on it, but we're not sure what's causing the issue yet.
  11. ~some chaos later~ 2017-12-20: Well, the forums are upgraded. I started the day trying to get my dev environment to display the admin control panel. It had insisted that it was going to, come hell or high water or the three switches I flipped - force an https link. I didn't have https set up in the dev environment, so it didn't work. It turned out to be surprisingly simple though - I uncommented two lines in the Nginx config file, added an exception in Firefox, and the page loaded like a charm. And I was worried! @Feld0 renewed our ipBoard license, and a few addon licenses, and I upgraded all the things. This broke some of the things, as you may have read about. The core forum upgrade took about 40 minutes. Other than scaring our ops team when mlpf went down for the upgrade, things went quite smoothly. It took about 40 minutes total, and another 5-6 hours to deal with the aftermath. The forum had some glitches which were quite worrying to see, although I think I've patched things up fairly well now with some horrible, horrible CSS. I was assuming the upgrade would be quieter, so I actually ended up spending a lot of time telling people what was going on while it was still ongoing. One thing that did take us by surprise is that the new ip.board reaction system installed enabled. We had assumed we'd have to check a box somewhere to enable it when we were ready. Work on the reactions system, probably to replace the brohoof system with something more meaty, is ongoing. I figure that if no major bugs crop up, the next thing to look at are the character pages. This is a bit of a long shot, however, as it's currently broken and I don't understand at all how that part of the system works. But nonetheless, worth a look.
  12. Ah. Thank you. Yes, that makes sense. Honestly? Maybe. 😬 I'm not that familiar with the forum software yet, but we'll see what can be done. You raise some good points. Thank you. One solution to the frown reactions problem we were considering was to simply remove the negative reactions entirely. It is a blunt yet effective approach, don't you think? ----- Anyway, I'm going to have to get some rest now. Catch y'all later!
  13. I think I've fixed the worst of it. (In horrible, horrible ways.) Let me know if anything else is obviously broken, please. I'll fix it soon, but I *need* to sleep now.
  14. @Lektra Bolt I don't think I was around back then; I don't know the formatting style you're thinking of. If it's gone, I doubt it's coming back - we're largely at the whims of IP.Board for that sort of thing, and I haven't heard of news around that. Sorry. It can be, now that we're back on a proper update track. I don't know what character pages are - I'm kind of new here, and my predecessor didn't leave very many notes I've been able to find. @VengefulStrudel Nothing too big. Some minor bugs should be fixed. The reaction system is getting a pony theme, and will be back. (tbh we assumed it would be off by default, and it enabling itself sort of took us by surprise.)
  15. @Nature Tune Could up attach the video? I'll try it here. ----- Yay! I'm glad to hear it's working, because I had absolutely no idea how to fix that one.
  16. That is… troublesome. Very troublesome. That absolutely shouldn't be. What browser are you using? What field are you trying to paste things in to? ---------------- Also, just a heads-up, but I'm going to update all our addons now because updated things are nice and hopefully not broken.
  17. Haha, yeah, I'm actually kind of digging the blue background myself. It makes everything more cheerful. We'll fix it back to where it was, if we decide like the colours we can put them back in right.
  18. Hey everyone, some early info on the forum upgrade. https://mlpforums.com/topic/173695-we-upgraded-the-forums-and-broke-things/ I thought it was going to be a lot quieter than it was. :P

    1. Snow

      Snow

      Do you think it is possible to trim the sides of that white bar up the top so it is inline with everything else? looks wise I think that is the only major problem, not really a bug or anything I know, but it does look rather untidy.
      123.thumb.png.7bb2969b540a8358d567914e11f42dd8.png

  19. Hello, everyone! We've just upgraded the forums to the latest and greatest version of IP.Board. This update has fixed several security issues, which we wanted; it has also introduced some new features, which we didn't want. For example, our likes system seems to have been replaced with some sort of reaction system, which we don't want either. We also need to fix our theme now. We hope to get it all sorted out quickly. If you notice anything broken, please let us know below. Thank you!
  20. Interesting! I had no idea about this - researching it lead me down an interesting little rabbit hole. (Mostly to the mysql documentation.) @Nightfall Gloam: Thanks!
  21. Well, after nearly two months of trying, I got my testing forum upgraded at last! This post is a little techier than I'd like because it's been written in haste and anger, but nonetheless — here is what I've been up to: 2017-11-22: Restored from a (thankfully!) recent backup, taken on 2017-10-31. Reapplied the steps I'd taken, which didn't take too long because I know what to do now. I did, however, run out of HDD space before the dump completed this time which was odd because it worked last time with 10gb to spare. Nonetheless, after I deleted my VM snapshots and resized the drive, everything ran begrudgingly to completion. I figured out how to get it to print errors this time too, which was helpful. There were a lot of them. 2017-11-24: After the DB finished loading into my VM yet again, I took a backup of the VM. On a separate disk drive. I then figured out how to load the production .sql files that make up the forum in such a way that they actually run in my VM. (For the record, I had to copy half the folders over by 'replace existing files', and the other half by 'keep existing files'. ) At this point, the admin panel started insisting that it wanted to run on https://, as opposed to http://, which my dev environment doesn't support at all. But I can sort of hack around it for now, so it's fine. I think most of my issues with the database have been caused by trying to take data from MariaDB (which mlpf runs on) and put it in to MySQL (which I installed on my dev VM). I can't figure out for the life of me how to get MariaDB running on an Ubuntu instance that's had MySQL installed on it - MySQL somehow poisons the box and prevents MariaDB from running, even after MySQL has been entirely removed. I tried working around that, but I'm the only person with this problem according to my google-fu. I also started setting up a new VM, one with MariaDB on it from the get-go, but I didn't get too far with it as I figured out the problems with the existing VM. Upon running the forum upgrader in my dev machine, I found that 17 tables had failed to import from production. They used a different database storage engine than I had. I modified the command I'd originally used to grab the production database to only grab those 17 tables. Running this command, I promptly took the forum down for 20 minutes while I downloaded 2.5 gigs of caching data. I'm so sorry if I caused anyone to loose a post they were working on. 2017-11-25: Carrying on. The data I now had let me run the upgrader a little further. Apparently I was missing two tables, which prevented it from running to completion. I recreated them and left them empty, since they didn't seem important. During the discovery process of what was wrong, I got this gem of a self-contradictory error message out of MySQL: mysql> CREATE TABLE `mlp_core_search_index` (`index_class` varchar(255)); ERROR 1146 (42S02): Table 'ipboard.mlp_core_search_index' doesn't exist mysql> It is complaining that it can't create the table because the table it's trying to create doesn't exist, which makes 0 sense. However, I could and did DROP it, so that solved that mystery. On with the upgrade! And, it runs to completion. But… were those tables important? 2017-11-26: I grabbed the data of the two missing tables from production, using more or less the same command that killed the forums on the 24th. (I am either brave or stupid.) After applying them, I ran the upgrade again, and again it worked without issue. One amusing part of the upgrade chain is that it asks you for a license key. I don't have one. But… the upgrader URL scheme is mighty predictable, and it turns out you can totally type in the URL of the next page without entering a key. I spent the rest of the day trying to get the admin panel to not require https, since I only have http on the test VM. Despite setting it in three or four places to only use http, it still forces https. It looks like I'm going to have to figure out how to self-sign a certificate so I can have https in the test VM. I'll probably have time to do that later next week, so that'll be good. Then I can continue on the upgrade path on my test VM until I hit the blocker that has prevented mlpf from upgrading. Then I will fix the blocker and upgrade mlpf.
  22. DDR

    hi again

    Hm, 'fraid this is beyond my very limited skills right now. Maybe try asking around on the Discord? It's pretty active these days, a good way to get people's attention.
  23. Oh, that's a good idea. If I can't resize the existing drive I'll do that. Thanks!
  24. The last few days here have been quite frustrating. 2017-11-09: Tried loading the database dump I downloaded on 2107-11-02. It didn't work, I think because it was dumped from a MariaDB database and loaded into a MySQL database. The dump was made to work by removing all occurrences of "PAGE_CHECKSUM=0" and "TRANSACTIONAL=0". This took a while because the dump is 15GB. After that was fixed, the dump was loaded into my testing VM's database in just over four hours. I think the two things I had to remove were optimizations. Disk space in the VM is a little tight now, so I uninstalled LibreOffice to free up some room. It seems I configured the VM with a 40gb virtual hard drive and the database now takes 30gb of it. The database loaded, I then blew a few minutes trying to figure out why I couldn't log in. It turns out having loaded the database from this forum I also loaded my password from this forum. The first thing that happens with the production DB loaded is IPS, our forum software, trying to write somewhere it doesn't have permission to. I made the installation folder, /var/www/, world-writable and world-executable. It works, but I feel all squidgy inside. We seem to be missing quite a few files, such as the theme, so most of the site doesn't work. The admin panel is a little more resilient, thankfully, and seems to be working fine - except for upgrade step 3, which throws a theming-related error. I figure I'll grab the likely files from the production server, there's a bunch in the web directory that seem likely candidates. However, I can't find the SSH details to access production, so I call it quits for the night here. 2017-11-10: Feld politely reminded me where the server was, and I found my notes by searching for the address. I copied the production files over to my local machine. Copying the production files over to the development folder didn't work outright, as a 404 error was thrown for the admin panel. I fixed some permissions, and traced the error down to something forcing HTTPS. While quite reasonable, HTTPS doesn't work on my dev machine. I eventually fixed that by coping the first half of the folders from production over using 'merge folders and overwrite existing files', and the second half using 'merge folders and keep existing files'. It's strange, and I don't understand why it works, but it works so yay! We have admin panel again, and now the theming-related error from yesterday is fixed. Step 3 out of the way, step 4 promptly fails when it can't download a file. Luckily, it comes with a link, so I download and apply the patch it wants. This fails as a blank page until I fix the permissions again. Now we are on to the real meat of the upgrade! First, the "converter" application can't be upgraded. After some asking around, it turns out it comes from - not the applications area on Invision's website - but bundled with the download of the forum software. I've decided to backburner installing the app for now, since we don't really use it, but I'll swing around and install it again after the upgrades are complete. The first step of the upgrade proper is to run a database command. It modifies a large table, so I need to run it manually. I can't run it, however, since the database says I'm out of virtual disk space. (Well, it said something completely unrelated, but I typed the error into Google and Google said 'disk space'.) So, since I can't seem to free up enough space to run the command in, I need to make the virtual disk itself larger. To do that, I need to make a backup of it first in case something goes wrong. To do that, I need more actual disk space to work with - my solid state drive has filled up. I had originally set my computer up so all the virtual disk images were on my larger hard disk drive, but it turns out that the mechanism for that had got broken. I copied over the testing VM's virtual disk to the HDD. I then set VirtualBox to use the new copy on the HDD, and tested that it worked there. I then deleted the old copy on the SSD to free up room, and got on with the process of enlarging the virtual disk. I ran on the HDD: > cp Poniverse-MLP-Forums.vdi Poniverse-MLP-Forums.vdi.bak cp: error reading 'Poniverse-MLP-Forums.vdi': Input/output error fiddle fiddle fiddle > rm Poniverse-MLP-Forums.vdi.bak -r rm: cannot remove 'Poniverse-MLP-Forums.vdi.bak': Read-only file system … … … I broke it. 2017-11-11: All the files that told the computer how to start were on the HDD, and the HDD is now dead as a doornail. After 3-odd hours spent trying to install Grub and mark the SSD as "a thing you can boot", I'm finally been able to start the computer again. I have been able to recover a backup of the virtual machine I took at the start of the month, since the current copy died with the HDD. My next step will be to redo the work I did so far this month based on the backup. It should be fairly quick, as I took good notes. But… still. It's been a fair chunk of time I've had to spend getting everything working again. I'll probably be able to take another crack at the upgrade on the 21st or 22nd. I should be able to borrow a replacement HDD before then, too, so that'll be good.
  25. I'm much closer to being able to test a potential fix now, but I still haven't been able to actually develop one. Hopefully we'll get to have a go at that next week. 2017-11-01: Installed the rest of the addons/plugins we use. Updated Ubuntu in the testing VM and restarted. This broke my shared folder (which I use to get files to/from the testing VM). I reinstated the folder with some difficulty and marked it non-transient, so it should stay around when I shut down the VM now. While that was in progress, I started downloading a dump of the forum database. 2017-11-02: Finished downloading the database dump. (All your secrets are mine now, muhahaha! Nothing can stop me now!) Next step: load it into the VM's database so I can roll it back on demand while I figure out why upgrading is taking so long. —————————— Thanks! I certainly need it these days, everything seems to be falling apart on me. (Except my Factorio game, for once. It's ran without the trains deadlocking for nearly 16 hours now! )
×
×
  • Create New...