Earlier today I noticed that my witness node had crashed.
I have not seen the issue that caused the crash before, but I will report on it on github soon.
When I noticed the problem, my witness node had already missed 2 blocks.
To disable a witness and get it off the witness schedule the operator needs to make a witness update with a new Signing Key:
Since I had no backup server, this is what I had to do.
The cli_wallet that comes with steem (https://github.com/steemit/steem/tree/master/programs/cli_wallet) has been broken ever since I have tried it 2 years ago. It times out within seconds and unless you have all the commands ready for copy + paste, you will not be able to finalize a single transaction with it.
Apparently, the issue still exists: https://github.com/steemit/steem/issues/2474
Usually, I do not bother trying, but the machine that I use for
conductor was broken, too. (this and my node fail are not steem's fault)
So I tried a dockerized cli wallet from steemit/steem.
So I tried installing
Conductor on a new machine, since the fast version did not work out. I need the comfort version anyways in the long run, right ?
After following all instructions to install
steem-python on a new Ubuntu 17.10, I try
steempy and get a
command not found as a return.
After fiddling for another 30 minutes and cloning into the repos manually, I can finally get it to work.
Only to encounter this issue:
However, steempy works and I can finally shut down my witness.
This took me way too long.
I know, I was ill prepared. I know, I could have done this much faster.
The fact, that not a single steem software works fresh out of the box is ridiculous, though.
Some of these issues are just sitting there since forever.
After I have now successfully shut down my witness, I run into the next issue:
This happened for the last year. Or pretty much ever since I run for witness.
To make it short:
When I turn the witness off, and try to turn it back on, it's broken
Most of the times, I have to fully resync and can not just start from where I stopped.
Even after a clean exit.
So ... The witness is resyncing now.
When I import a block_log from somewhere I have a 30% chance of it working and not failing to reindex. Resync takes even longer than 3 days, but I can not be bothered to put any more effort into this shit right now.
My witness will be up again in a few days.