You are viewing a single comment's thread from:
RE: 10 Things I Did While The Blockchain Was Down
After I did most of the things on your list, I watched the meltdowns in the steemit chatrooms of witness and witness-social. I see a slight problem with the TOP 20 witness things, I guess it is hard to even get 20 people to see that the blockchain had stopped working. Ran into a wall. But it was fun to watch them in the chat, I guess there is an ultra secret slack chat room somewhere the "real" witnesses talk.
They need to clean the witness table up also.
number 14 -Witness Disabled
This witness does not have a signing key either at the owners request or because too many blocks have been missed.
number 26 and 31 same thing. Number 35 mr bainfeld same thing.
The witness list needs purging.
But like I said it was fun watching both rooms.
Hey, @bashadow.
I was headed over to Steemit.chat on Asher's recommendation, but I couldn't get in. Not sure now what the problem was. Password, something, which I don't remember having an issue with the first three times I tried it seven months ago, when I swore it off.
I guess I would have enjoyed it, too, to a point.
I'm afraid many of the top 20 witnesses aren't regularly active on STEEM. They've got their own projects and they're busy with them. We've got people calling for more dev witnesses now. Well, okay, but it wasn't the witnesses who came up with the HF 20 code, to my knowledge. And I don't know if it's the witnesses' fault or primary blame that it went down, either.
I'd prefer they not rubberstamp anything. It would be good if there were among the witnesses maybe four or five pairs of eyes to look through it, but what do you look for? I don't know. I suppose there's a need to know how to get things up and rolling quickly when things go sideways, too, but that really shouldn't amount to much more than running the previous good version and up we go.
It's a little discouraging to me given what I'm told is the sheer magnitude of scope. The only thing that can be done, I'm repeatedly told, is what happened. Wait until it breaks. Fix as fast you can. Wait for the next one.
I agree totally with the witness list. There's a reasonable time in which a witness could be down, but weeks isn't it. And a witness should, on their own, advise what they're doing, too. But there's no reason why their should be a witness list and an active list, especially since the witnesses are being compensated based on one or the other.
From what I saw in the chat, it seems the bug, or one of them was a small difference between if I remember right 12 seconds verse 12 hours. a misplaced decimal point or something. It was working fine on the testnet, or so it seemed to timcliff, and then the bug was found by an action, and boom kapowee and the lockdown invalid action or whatever.
The thing people need to remember is that the safety system did work. It stopped and waited to be fixed. My guestimate 12-13 hours of downtime. Not sure on the exact number but no funds lost everything safe and sound and back to normal. For a software bug mind you.
MS, Apple, and other software people hide their bugs people don't know their stuff is not safe until after MS, Apple, etc tell them or they get caught by the media, "oh yeah, we had an issue with that a few months ago, it tooks us 6 weeks to fix it but your all safe now". type statements. So people missed a few hours of time, that in the end is all they lost.
I can't speak for anyone else, but my issue isn't with what comes after the failsafe kicks in, it's before. It's what can be done to prevent these things from happening. So far, the answer is nothing. We just wait until it breaks. I don't know how that compares to those other companies. I can't speak to the transparency of Steemit Inc, either, when it's the entire blockchain that goes down. Rather obvious something happened, and it's affecting everyone trying to get on at once.
I was thinking it was basically 11 hours.
The reason why it was working on the testnet is because the testnet didn't include testing for what happened, which is odd, since what happened was a result of the change in upvoting in the current 12 hour lockout period. So, the bug makes it through.