Did we save the world yesterday? Maybe, perhaps, who knows...

erkan (60)in #gridcoin • 7 years ago (edited)

Below is a brief recapture of what happened yesterday (a dev asked me to make a bug report, but let's gather first all info...)

Prelude:

Forks happen all the time, you could say with every new block. With every cryptocoin that is out there, and the clients notice this soon and bring you back on track.
In general all the clients agree themselves and make a "consensus" which chain will be the "correct one", from their POV.

So, how do you actually know that something is "wrong" in your Gridcoin client?

You suddenly stake too fast blocks, or you see a red out-of-sync notice, ...
See more here.

So, what to do when you notice you are on a fork?

check your block hash with fediverse on IRC (use the !blocks and !verify commands): if your current block's hash (e.g. for #958826) is wrong, please also check 10 blocks past (e.g. #958816). If 958816 and upper blocks are fine, likely your client will self-heal and you can enjoy your cup of coffee
If you notice though, that the hashes are wrong for a longer series: please restart your client (your client will try to fix your situation at start)
If this still does NOT help: ask first in IRC what to do (the most likely answer will be to use the !snapshot or !bootstrap).

The situation today is GOOD (and also was good yesterday after like 1-2h):

we have >70 million net weight
gridcoinstats is special and OK (it may swing from time to time to past blocks but recovers soon enough to the correct chain)
asia is still lagging, but it's on the correct chain
holy transaction is still way in the past

What we can conclude here already is:

a few more public explorers (and nodes) wouldn't hurt, especially since the admins of the current block explorers are lazy... only the admin of gridcoinstats was quick to reply and act in chat. If you have free time, why don't you set up one also ? fediverse uses the current public explorers for its DPP (disaster prevention plan) feature

So, what happened yesterday?

Yesterday, around 17:30 PM UTC+2 we noticed block explorers were on a fork/out-of-sync/lagging and also several users reported such...
and the netweight around that time in my Gridcoin client dropped to 44.6 million (the last few days net weight was in the ranges of 70-100 million)

17:33:

me: let me quickly verify the overall situation on the block explorers and such
me: Holy transaction still fighting they are at: Thu, 29 Sep 2016 (they have some problems since a few weeks now)
me: yes, gridcoinstats is off ( has different hash)
me: Asia is OK, but lags

The net weight fell down more to 42.9 million

while the "other" fork/chain was at like 13-17.2 million

I sent out a tweet also.

At 6:20 PM we reached again >60 million netweight which was a good sign (see (A)).

So, did we save the world yesterday ?

Maybe, perhaps, who knows... but coordinating in IRC and slack that users get on the right track helps prevent that a disaster will happen (A). At the last major fork event in March we contacted exchanges/services to halt Gridcoin, and one effect of that was that users could not deposit/withdraw e.g. on Poloniex for several weeks (also fault of Poloniex for not reacting on support tickets)
Also, keep this in mind: normally clients handle consensus themselves... Actions like telling users to restart their clients "influences"... If evil-minded persons with enough planning beforehand do such a thing: it can lead to manipulation

(A)

from irc (there are more factors, but for now this explanation is "good enough"):

me: the weight is not so the factor

me: it's like a virus

me: the connections determine it mostly

me: if you have bad neighbours you won't ever recover

me: from the weight we see if it is same net weight as before it was known

me: that gives then a more trustful feeling

me: see also:

~~me: and I may note that this was CLOSED as won't fix (I disagree~~)

me: so basically, we had before: 70-80 million netweight