Author: almindor

Server issues

Server issues

UPDATE: 2018-05-25

Geth has been identified as the ultimate culprit by eating all available RAM causing the server to go to a crawl. I’ve switched the node to parity to fix the issue.

The change to parity as the main websocket thin client provider however has caused an error related to filters uninstall call in the RPC. Etherwall 2.2.3 has been released with a fix so parity can be used as the backend.

Syncing is still in progress. I will update the server status and remove the warning message when it’s done.

ORIGINAL: 2018-05-23

The main Etherwall server hosting both the services for thin client and this website is experiencing memory related issues atm.

The culprit seems to be geth itself taking up all RAM (32gb+) and causing OOM killer to kick in. The last long downtime might have been related to this issue as well.

Last server downtime caused a loss of recent DB data as well due to a hard reset. I’m currently trying to find out how to limit impact if geth decides to gobble up all the RAM again.

The DB data lost is related to the historic transactions requests. The DB is being rebuilt from scratch but historical transactions list will remain outdated for a few days.

 

Server down

Server down

The Etherwall server went down sometime around 13th May. I am currently out of country with no way of logging in.

I’m returning on Friday 18th and should be able to restart the service by then.

If you need to move your assets and use the thin client mode (default) here are your best options for time being:

  1. Switch to full node in settings and let geth synchronize on local machine (requires quite a lot of disk space and bandwith)
  2. Use another wallet and import the geth keys. You can export individual address geth keys from etherwall (if you disable thin client) or access them directly in the datadir folder

I’m sorry for this extended downtime but there’s nothing more I can do at this point. I’ll try to ensure that I always carry the ssh keys with me next time I travel out.

Server upgrade

Server upgrade

Ever since cryptokitties started spamming the network the geth node stopped keeping up due to IO slowness. There’s a planned server upgrade adding a faster SSD drive for chaindata storage at 19:00 CET. This should hopefully fix the synchronization issues.

NOTE: there will be a full blackout of the server including this page for a short time followed by some time when the thin client won’t be available as the data is moved over to the faster disk. Updates will be posted here as the upgrade is performed.

 

Cryptokitties performance test

Cryptokitties performance test

Cryptokitties has landed on ethereum and cause major havoc by overloading the network. In my opinion this is a good thing even tho it gives me headaches right now because it caused Etherwall’s remote node to be behind on synchronization.

Our server runs on a big but slower disk now and this caused it to slow down on accepting new blocks when we approached the cryptokitty territory. There are just too many transactions to check. I’m keeping an eye on the server if it can’t catch up I will try and resync from scratch using the pruning method.

If there are those of you who could donate a server please consider it so we have broader coverage for the thin client. Etherwall is 100% voluntary with no monetary gain.

ERC20 support is here

ERC20 support is here

Etherwall v2.1.0 is about to be released with ERC20 support!

This brings Etherwall to feature parity with the official ethereum wallet with the nice bonus of a thin client so people don’t have to download 40gb of chaindata.

Some performance improvements have been added in this release as well.

Last minute fixes

Last minute fixes

Ethdev has done it again and has released a quick last minute fix for go-ethereum just before the Byzantium hard fork.

I’ve waited patiently for 1.7.1 knowing that 1.7.0 would soon be followed by an update, but I didn’t expect them to make another rushed hushed fix just hours before the big fork.

Seems EthDev is hasty as usual.

Etherwall v2.0.8 has been released with geth 1.7.2 for windows and mac and a version check that errors out on older versions to ensure everyone is running a DDOS resistant version of geth.

Server warnings

Server warnings

Version 2.0.5 of Etherwall adds support of showing server warnings.

This is required as the main geth node server is running out of space again and needs to be resynced and pruned, which means downtime.

I’m also changing the server-side to provide node endpoints so that the clients can connect to different remote nodes once available. If anyone has a server they can donate to Etherwall please provide the details at this github issue.

Version 2.0.6 was released with a quick fix to a full-node initialization bug.

Quick fix

Quick fix

Version 2.0.3 went through a refactor and I completely missed a bug in contract/events.

Version 2.0.4 fixes that alongside bringing initial part of ERC20 support. Note that this is a WIP and the only implemented part is discovery of token contracts and getting the symbol back.

Now that eth_call is implemented however, you can use Token contracts “manually” until v2.1.0 is released with full support.

Thin client downtime

Thin client downtime

UPDATE: 2017-08-15 the server has been restored

The server hosting geth for the “thin client” mode has ran out of disk space causing Etherwall to fall behind for people using the thin client.

I have restarted the server to re-sync from scratch pruning the db to keep the size down. This means a downtime of up to one day when using the thin client.

Don’t panic if you’re using the thin client mode and see zeroes in your accounts. Until the synchronization is done on the server your account balances will be zero or may contain old data.

I’ll add a better way to inform about the server status from within Etherwall itself in the next release.

Long Comming!

Long Comming!

Etherwall version 2.0.0 has been finally released!

This version brings a lot of exciting changes:

  1. Thin client support
  2. TREZOR ONE hardware wallet support
  3. Rinkeby testnet

Thin client means no more cumbersome downloads of 30GB+ for the ethereum blockchain! Anyone who wants to run a full node can do so of course but for most people this will be a great improvement saving both disk space and improving performance as Geth will no longer need to check all the blocks on your machine. For more info on the thin client mode see FAQ.

TREZOR ONE support, long on the wishlist, is now a reality. Just plug in your TREZOR ONE and go from there. For more info on TREZOR ONE support see FAQ.

The ethereum testnet has been very unstable. This release changes the testnet chain to rinkeby hopefully making it more useful. NOTE: testnet is only supported in full node mode.

This is a major version upgrade so everyone will be prompted to review settings on first start.