(Topic ID: 291292)

Back after a major outage

By robin

3 years ago


Topic Heartbeat

Topic Stats

  • 331 posts
  • 210 Pinsiders participating
  • Latest reply 2 years ago by boagman
  • Topic is favorited by 6 Pinsiders

You

Linked Games

No games have been linked to this topic.

    Topic Gallery

    View topic image gallery

    Pimball.GIF
    image (resized).jpg
    pasted_image (resized).png
    69DFCFDC-E12D-4C2D-8036-6C15CDB4799D (resized).png
    waterdatacenter (resized).png
    uOcsyp (resized).png
    B4EC046D-B13A-4FFE-8F23-3C245F7AE44F (resized).png
    umm-humm-when-keeping-it-real-goes-wrong (resized).jpg
    10b05741d4aa7afec9de5412a443548c (resized).jpeg
    49E5334B-EE51-4A76-B09B-8E99668037D7 (resized).jpeg
    giphy.gif
    pasted_image (resized).png
    6DB69A6E-3813-42D2-91AE-969F14A01757 (resized).jpeg
    pasted_image (resized).png

    You're currently viewing posts by Pinsider robin.
    Click here to go back to viewing the entire thread.

    220
    #1 3 years ago

    We're back!

    What a crazy week it has been for Pinside (and a stressful week for me, I might add) with our first major outage in the 20 years or so that Pinside exists. Sure, we've had hiccups but nothing like this. The cause: a power surge, followed by a fire in a generator, followed by sprinklers and our server going down. I wasn't there to see it, but I don't think it was pretty.

    Anyways, we're running on a temporary server now, hosted all the way over in Germany (well, that's close for me). Please be nice to it, it's not as powerful as our previous one and there may be problems. Also, not everything functions yet: We have no e-mail notifications yet. And no maps. And probably a bunch of other stuff that doesn't work right now. But at least we're back. I'll be moving to a brand new server (and a different hosting company) later on but I don't want to rush that choice.

    As for our data recovery procedure. Well there really wasn't one in place as you may have noticed. Sure I had dual power supplies, dual network uplinks, dual (raid) drives. But I would not have expected a datacenter fire. I should have. I should have had a backup server somewhere, ready to spin up and continue service after a few hours. Truth is, I never got around to setting this up. But I will obviously be working on this, so a downtime like this cannot happen again in the future.

    My last database backup was dated 10:02pm, while the server went down 11:39pm. So we lost 1h37m of data. That means any posts, PMs or Marketplace ads placed during that time are gone. And even if they restore our server (they're still working on it but are now giving me an ETA of "weeks") I will not be bringing that data back as that would be a technical nightmare.

    If you made a donation or marketplace payment, please contact me. I will also be proactively look into any payments made during that 1h37m and take action or contact you.

    Speaking of donations. Feel free to make a donation, if you're happy that we're back!

    Let's get the pinball talk going again and please bear with me while I try to keep this temporary server running nicely.

    -- Robin

    PS: Seeing any issues? Let me know!

    65
    #5 3 years ago

    Well, you all didn't take long to crash our temporary server. After 30-40 minutes it went down smoking. So I went back to the drawing board and ordered a brand new server. This should work, but please bear with me as I will be tweaking and tuning some more over the next week(s). The upside is that the new server is twice as powerful as the previous one, so with the right approach this could all end better than it started.

    Anyways, if you're able to read this: welcome back!

    52
    #232 3 years ago

    Hey everyone. Thanks a million for all the nice words in this thread. Believe me, it's much appreciated!

    Last week was really a nightmare scenario. Having moved over to a smaller Web Hosting company last year with the idea of getting more direct attention in case of network or hardware issues didn't work out quite like I had hoped.

    So what exactly took me so damn long to get back online? I had backups right? Well, for anyone who's interested in that story, I'll try to recall at some time - but let's just say I wasn't prepared for this kind of disaster. Redundancy is cool, until your datacenter catches fire.

    But of course I should have prepared for this, in hindsight. And I will be in the future as I'm working on a good DR strategy as we speak.

    ... waterdatacenter (resized).pngwaterdatacenter (resized).png

    So what's cooking now?

    I'm still working on getting everything back up. I got email fully working again today. Of course, being a new server and IP, Outlook immediately threw us on their blacklist. A few emails with their team and several hours later they mitigated this.

    Got email replies working again!

    Just got the Pinside maps server back online!

    Currently working on Marketwatch emails.

    And of course working on backup strategies.

    Coming back soon: topic image gallery, Pinside achievements and anything else that is temporarily disabled or that I forgot. Not much else, I think? Let me know what you are missing besides the stuff I just mentioned.

    Finally, a huge thanks to people who (offered) help/support. And my wife for being extremely patient with me as I was away most of the week and probably terrible to be around during the times I was at home. My brother Martijn of course for moral support. Forceflow. Jay@Scorbit, Jeff@TWIP for relaying messages that we didn't disappear to Timbuktu with all your gif uploads. And Ed R for being funny.

    ** And all of you who donated in the past 48 hours. Amazing to see all that support **

    To be continued. good to be back!

    11
    #240 3 years ago

    As of now, Pinside MarketWatch™ notifications should be working again!

    22
    #245 3 years ago
    Quoted from Isochronic_Frost:

    I shudder to think about losing all the precious resources and stories, vid’s guides, and especially all the archives of pinheads who have coined up that great pinball hall in the sky.
    Their stories are maintained here and nowhere else, they’d never come back if we lost them.

    You're absolutely right. You'll be pleased to hear that, going forward, Pinside is now running *hourly* data backups. These get stored in 3 separate datacenters by 3 different companies in 3 different geographic locations. Encrypted and safe, also from ransomware due to object locks.

    And to give you another Shudder story: As I was restoring backups to my temporary server, I swear I could not remember where I had stored my data encryption key. Maybe it was the stress or what not, but it took me an hour of frantic searching. After having a coffee, it dawned on me where it was. I literally shouted out a relieved "Whoohoo!" when I found it. Another learning moment, ain't gonna happen to me again.

    36
    #270 3 years ago

    Super quick update. All is going great with the new server!

    The Pinside Top 100 is back to weekly updates.

    I’m still working on the image server.

    And don’t worry: Everything that was on Pinside before the outage will be back. I mean achievements and topic image gallery. I just need some more time while I first move to a new image system. More on that later!

    Good to be back!

    #281 2 years ago
    Quoted from Jappie:

    Great work, Robin!
    Is it correct that adding a video to the Multimedia section of a Splash page isn't working yet? I'm getting an error when I try.

    This issue should be fixed, Jasper.

    “Posts by you” is a known issue for which I need some more time to get it optimized and working properly again.

    Collection page bug is news to me, I’ll see if I can replicate it here.

    #295 2 years ago

    Looking into it. The files are still present in backups so something weird's going on there...

    Found the cause of some images not working: the resolution was higher that I would allow. This has been fixed but you may need to clear your cache to see it.

    Quoted from Lovef2k:

    I tried several time to start my PS store but I keep getting an address error even though I re-entered my address several times and checked all boxes.

    Could you contact me via PinsideMail and possibly include screenshots of what you are doing (and of the exact error message)? Thanks!

    Quoted from Rodent:

    I need to dig further, but I am experiencing an issue with adding a new high score:
    ...
    interesting note: after modifying the score to add the image, my Pulse notes that editing a high score '//not actually possible at the moment' but the edit is completing

    Seeing the same thing here, I will investigate why this is happening. Meanwhile, I found that tapping the score box will show the keyboard. So the workaround is to enter the score first, then tap the machine name field to enter the machine.

    The 'not actually possible at the moment' is from when editing scores was not yet possible. I forgot to remove that but will do so

    #299 2 years ago
    Quoted from Pinball_Hawk:

    robin the date markers are broken for fourm posts.
    [quoted image]

    Whoah!! What happened to your fonts?

    What do you mean by date markers? Do you mean the green "unread posts" boxes? Maybe log out and back in again.

    Quoted from Rodent:

    I want to submit a new score because the 'keyboard' won't pop up

    This should be fixed now!

    #312 2 years ago

    Interesting findings. I always figured Linux would be fine falling back to Open Sans font.

    Maybe Pinside should use font embedding for our main font? Currently our font stack is font-family:"Verdana","Open Sans",sans-serif;. Although I don't think I can embed Verdana without paying a licensing fee.

    Another idea could be to install the MS Core Fonts TTF package in your linux distro. It adds some basic fonts, including Verdana. This way you'll see Pinside as was intended.

    But that's all font talk. Your issue is with the colors and I wonder if you are doing some kind of override causing this?

    #315 2 years ago
    Quoted from Jediturtle:

    I also noticed the "all posts made by you" section under the Posts tab hasn't updated since before the crash. Just in case that one was flying under the radar. Thanks for all your hard work Robin!

    The "All posts made by you" problem should be fixed now. the page also loads really fast now, compared to the 3 sec loading time it took before (database query rewritten). Please take a look and let me see if returned results are as expected.

    #316 2 years ago
    Quoted from rkahr:

    robin - here's a post I made a year ago that may help exemplify a couple little things:
    https://pinside.com/pinball/forum/topic/the-crazy-mansion-by-the-pinball-amigos/page/194#post-5354640
    First, the Pinside Activity graphs that used to be next to the heart icon just beyond our user name isn't active. The pic in that post has the graph circled in red.

    Yeah, this is related to achievements and I hope to bring it back soon.

    Quoted from rkahr:

    Second, hovering over the picture reveals a button in the lower left that provides a "Direct Link to Full Size Image". Clicking that gives a NGINX 404 error in a new tab - the click directs to https://images.pinside.com/2/84/08/28408292b8fd18d7f728943360276588fd6f6e0c.png.
    Hope this is helpful for your continued diagnostic/recovery quest.

    Cool, thanks. I will fix that today.

    #320 2 years ago
    Quoted from Bryan_Kelly:

    Not sure if it's been mentioned, but after loading pics in a market place ad, I'm not able to rotate them.

    Thanks, I'm aware and it's on my to-do list!

    Quoted from Ericpinballfan:

    I'm still not able to load pictures on my Droid device.
    I tap it, it lets me select resolution size but nothing after that.

    The upload button is there, but it's tiny on some devices. Fix coming!

    Quoted from Mr_Tantrum:

    You may be aware, but animated GIFs seem to be taking extra long to load on the new server and often times won't load at all.

    Doing some research into this. Our new image server apparently increases file size on some gifs. I'm implementing some smart code today to always serve the smallest file (either processed or raw). Please also note that the first load of an animated gif (or any image resize for that matter) will require some (more) processing time. After that initial processing, the image is cached in our CDN and subsequent loads will be mucho faster for everyone else.

    #321 2 years ago

    Quick update:

    Image rotate should be functioning now. It isn't well tested, but should work. Please report any bugs!

    The blue "Add Image(s)" button should now always be visible, also on Android phones

    Gif images will now be automatically converted to WebP, if your browser accepts that format (Chrome, Firefox and the latest Safari on Big Sur). WebP gives a huge size reduction over gif! If your browser does not support WebP, I'm now leaving the gif image alone and serving it as uploaded (which could mean large file sizes!)

    Example:

    This animated image of my buddy Pim is 2,5 MB as a .gif file. But only 508 kB when served as WebP.

    Pimball.GIFPimball.GIF
    1 week later
    #329 2 years ago
    Quoted from EJS:

    Are the achievements thingys coming back? Or did I overlook that in a recent update?

    They are, just not high priority right now. Sorry!

    You're currently viewing posts by Pinsider robin.
    Click here to go back to viewing the entire thread.

    Reply

    Wanna join the discussion? Please sign in to reply to this topic.

    Hey there! Welcome to Pinside!

    Donate to Pinside

    Great to see you're enjoying Pinside! Did you know Pinside is able to run without any 3rd-party banners or ads, thanks to the support from our visitors? Please consider a donation to Pinside and get anext to your username to show for it! Or better yet, subscribe to Pinside+!


    This page was printed from https://pinside.com/pinball/forum/topic/back-after-a-major-outage?tu=robin and we tried optimising it for printing. Some page elements may have been deliberately hidden.

    Scan the QR code on the left to jump to the URL this document was printed from.