|
Title: Version 0.6.0 released Post by: Gavin Andresen on March 30, 2012, 03:19:38 PM Bitcoin version 0.6.0 is now available for download at:
http://sourceforge.net/projects/bitcoin/files/Bitcoin/bitcoin-0.6.0/test/ This release includes more than 20 language localizations. More translations are welcome; join the project at Transifex to help: https://www.transifex.net/projects/p/bitcoin/ Please report bugs using the issue tracker at github: https://github.com/bitcoin/bitcoin/issues Project source code is hosted at github; we are no longer distributing .tar.gz files here, you can get them directly from github: https://github.com/bitcoin/bitcoin/tarball/v0.6.0 # .tar.gz https://github.com/bitcoin/bitcoin/zipball/v0.6.0 # .zip For Ubuntu users, there is a ppa maintained by Matt Corallo which you can add to your system so that it will automatically keep bitcoin up-to-date. Just type sudo apt-add-repository ppa:bitcoin/bitcoin in your terminal, then install the bitcoin-qt package. KNOWN ISSUES ------------ Shutting down while synchronizing with the network (downloading the blockchain) can take more than a minute, because database writes are queued to speed up download time. NEW FEATURES SINCE BITCOIN VERSION 0.5 -------------------------------------- Initial network synchronization should be much faster (one or two hours on a typical machine instead of ten or more hours). Backup Wallet menu option. Bitcoin-Qt can display and save QR codes for sending and receiving addresses. New context menu on addresses to copy/edit/delete them. New Sign Message dialog that allows you to prove that you own a bitcoin address by creating a digital signature. New wallets created with this version will use 33-byte 'compressed' public keys instead of 65-byte public keys, resulting in smaller transactions and less traffic on the bitcoin network. The shorter keys are already supported by the network but wallet.dat files containing short keys are not compatible with earlier versions of Bitcoin-Qt/bitcoind. New command-line argument -blocknotify=<command> that will spawn a shell process to run <command> when a new block is accepted. New command-line argument -splash=0 to disable Bitcoin-Qt's initial splash screen validateaddress JSON-RPC api command output includes two new fields for addresses in the wallet: pubkey : hexadecimal public key iscompressed : true if pubkey is a short 33-byte key New JSON-RPC api commands for dumping/importing private keys from the wallet (dumprivkey, importprivkey). New JSON-RPC api command for getting information about blocks (getblock, getblockhash). New JSON-RPC api command (getmininginfo) for getting extra information related to mining. The getinfo JSON-RPC command no longer includes mining-related information (generate/genproclimit/hashespersec). NOTABLE CHANGES --------------- BIP30 implemented (security fix for an attack involving duplicate "coinbase transactions"). The -nolisten, -noupnp and -nodnsseed command-line options were renamed to -listen, -upnp and -dnsseed, with a default value of 1. The old names are still supported for compatibility (so specifying -nolisten is automatically interpreted as -listen=0; every boolean argument can now be specified as either -foo or -nofoo). The -noirc command-line options was renamed to -irc, with a default value of 0. Run -irc=1 to get the old behavior. Three fill-up-available-memory denial-of-service attacks were fixed. NOT YET IMPLEMENTED FEATURES ---------------------------- Support for clicking on bitcoin: URIs and opening/launching Bitcoin-Qt is available only on Linux, and only if you configure your desktop to launch Bitcoin-Qt. All platforms support dragging and dropping bitcoin: URIs onto the Bitcoin-Qt window to start payment. PRELIMINARY SUPPORT FOR MULTISIGNATURE TRANSACTIONS --------------------------------------------------- This release has preliminary support for multisignature transactions-- transactions that require authorization from more than one person or device before they will be accepted by the bitcoin network. Prior to this release, multisignature transactions were considered 'non-standard' and were ignored; with this release multisignature transactions are considered standard and will start to be relayed and accepted into blocks. It is expected that future releases of Bitcoin-Qt will support the creation of multisignature transactions, once enough of the network has upgraded so relaying and validating them is robust. For this release, creation and testing of multisignature transactions is limited to the bitcoin test network using the "addmultisigaddress" JSON-RPC api call. Short multisignature address support is included in this release, as specified in BIP 13 and BIP 16. Thanks to everybody who contributed to this release: Alex B Alistair Buxton Chris Moore Clark Gaebel Daniel Folkinshteyn Dylan Noblesmith Forrest Voight Gavin Andresen Gregory Maxwell Janne Pulkkinen Joel Kaartinen Lars Rasmusson Luke Dashjr Matt Corallo Michael Ford Michael Hendricks Nick Bosma Nils Schneider Philip Kaufmann Pierre Pronchery Pieter Wuille Rune K Svendsen Wladimir J. van der Laan coderrr p2k sje397 Special thanks to Sergio Lerner and Matt Corallo for bringing potential denial-of-service attacks to our attention. Title: Re: Version 0.6.0 released Post by: minimalB on March 30, 2012, 04:05:56 PM Thanks for your hard work!
Title: Re: Version 0.6.0 released Post by: finway on March 30, 2012, 04:48:56 PM Bitcoin version 0.6.0 is now available for download at: http://sourceforge.net/projects/bitcoin/files/Bitcoin/bitcoin-0.6.0/test/ http://sourceforge.net/projects/bitcoin/files/Bitcoin/bitcoin-0.6.0/ EDIT: Congratulations and thanks! I'll have a bottle of beer to cheer this version. Title: Re: Version 0.6.0 released Post by: occulta on March 30, 2012, 05:19:13 PM nice, keep it up ! ill be donating again
Title: Re: Version 0.6.0 released Post by: rjk on March 30, 2012, 05:20:06 PM The signmessage feature is nice - but is there a way to verify the message within the GUI?
Title: Re: Version 0.6.0 released Post by: Daily Anarchist on March 30, 2012, 05:31:29 PM When will the Ubuntu Package be updated?
Title: Re: Version 0.6.0 released Post by: etotheipi on March 30, 2012, 05:50:11 PM Btw, I would caution against saying "Signing messages ... to prove that you own an bitcoin address ...". It's very difficult for someone to say "prove you own this address" and then you can do so. A MITM could ask the person who really does own the address to prove they own it then forward the response packet to the original requestor as if they owned it themselves. There are ways to do it, but it is not likely to be executed properly, by hand.
I recommend promoting the feature as "Message signing, with address-based authentication" or something more PR-friendly. For instance, you need a shipping address to send the 1,000 BTC merchandise that was paid for [partly] with address X. You would instead ask for a shipping address signed by address X. The MITM wants to divert the package to his own location, but can't do so without breaking the signature. The important part is that the same person who owns X approves of the message that was being signed, without regards for where the person is or when they signed it. Perhaps that's already in your line of thinking, but I think it's a distinction worth making. Title: Re: Version 0.6.0 released Post by: MysteryMiner on March 30, 2012, 06:22:53 PM The address signing is counteranonymous and against initial goals of Bitcoin. QR codes are not used worldwide and will fall in disuse much faster than IrDA links did. I smell feature bloat.
Title: Re: Version 0.6.0 released Post by: etotheipi on March 30, 2012, 06:37:44 PM The address signing is counteranonymous and against initial goals of Bitcoin. QR codes are not used worldwide and will fall in disuse much faster than IrDA links did. I smell feature bloat. I'm not sure I follow how it is counteranonymous. You fund a website/service with 20 BTC coming from address X. You can then manage your account over the internet without ever creating a username/password, supplying name or email address. The only thing that matters to the web service is that the same person who originally sent them 20 BTC is the same person telling them to transfer it to another address, purchase something, buy into a poker game, etc. It has the same security and anonymity features as Bitcoin itself. For now, such messages might have to be manually constructed, but in the future, your BTC private key could be used to initiate SSL connections, etc. Title: Re: Version 0.6.0 released Post by: Serge on March 30, 2012, 07:32:37 PM nice, glad noirc made into default setting with this update.
i'll be updating soon from 0.4 Title: Re: Version 0.6.0 released Post by: istar on March 30, 2012, 08:13:02 PM Could not install on a Nordic windows7 32bit when logged in as a non admin, tried twice.
The third time I changed the installer directory to "program" instead of "program files" and it worked. However a casual computer user will having problem will not figure that out. If its possible to fix it would probably be good. Really great work guys! Title: Re: Version 0.6.0 released Post by: Tritonio on March 30, 2012, 09:25:57 PM The signmessage feature is nice - but is there a way to verify the message within the GUI? I'd also like to know this. Title: Re: Version 0.6.0 released Post by: bitlotto on March 30, 2012, 09:31:43 PM Does the software use https://en.bitcoin.it/wiki/BIP_0021 as the QR format?
Title: Re: Version 0.6.0 released Post by: Gavin Andresen on March 30, 2012, 10:26:46 PM When will the Ubuntu Package be updated? When the maintainer (Matt) has a little time.The signmessage feature is nice - but is there a way to verify the message within the GUI? No, not yet. Anybody know if there's a web page that will do the verification? (would be easy to create one...)Does the software use https://en.bitcoin.it/wiki/BIP_0021 as the QR format? Yes.Title: Re: Version 0.6.0 released Post by: Diapolo on March 30, 2012, 10:30:57 PM The address signing is counteranonymous and against initial goals of Bitcoin. QR codes are not used worldwide and will fall in disuse much faster than IrDA links did. I smell feature bloat. It's pretty easy to disable support for that one, during compile time ... so if it's a sucker it's not hard to remove support for it ;). @Gavin: The linked version has still the Beta flag in the version string, perhaps you forgot to re-label it. Dia Title: Re: Version 0.6.0 released Post by: etotheipi on March 30, 2012, 10:51:15 PM The signmessage feature is nice - but is there a way to verify the message within the GUI? No, not yet. Anybody know if there's a web page that will do the verification? (would be easy to create one...)Gavin, I put a sign&verify GUI into Armory, and used a BIP-10-like format to create and verify "signature blocks." Here is an example of the signature block and the user interface. (https://bitcointalk.org/index.php?topic=56424.msg776163#msg776163) The UI needs to be cleaned up a bit, but it does work. Unfortunately, I am not yet compatible with the Satoshi client and I suspect I'll need to handle compressed public keys to do so. But I want to get it compatible as soon as I can, so can you make sure that it is documented so I can match? You also might consider a similar "signature block" technique (or do you already have it?) Title: Re: Version 0.6.0 released Post by: Omni on March 30, 2012, 10:52:54 PM Gavin you're a legend.
Title: Re: Version 0.6.0 released Post by: MaxSan on March 30, 2012, 11:10:19 PM Thanks to Gavin all others for you continued hard work on the very core of bitcoin.
Title: Re: Version 0.6.0 released Post by: jgarzik on March 31, 2012, 02:21:13 AM Thanks to all our contributors and testers for their hard work.
Title: Re: Version 0.6.0 released Post by: niko on March 31, 2012, 03:38:35 AM False alarm - after 5 minutes or so it started going up. Title: Re: Version 0.6.0 released Post by: theymos on March 31, 2012, 04:38:29 AM @Gavin: The linked version has still the Beta flag in the version string, perhaps you forgot to re-label it. All Bitcoin releases are considered beta. Title: Re: Version 0.6.0 released Post by: gorgo1 on March 31, 2012, 10:23:35 AM Thanks for the updated version of the client Gavin.Installing now. :)
Title: Re: Version 0.6.0 released Post by: Jeremy West spendbitcoins.com on March 31, 2012, 10:36:14 AM Initial network synchronization should be much faster (one or two hours on a typical machine instead of ten or more hours). Here's something I've been wondering for awhile now, but only now remember when I'm actually on the forums... Would it make sense to include a version for new adopters to download which includes an up-to-date (at time of release) blkindex file rather than a blank blkindex file so that they are nearly synced at installation? Title: Re: Version 0.6.0 released Post by: slothbag on March 31, 2012, 10:49:50 AM Awesome work Gavin and team..
The fast block download is a huge improvement. Even my low powered atom PC managed to catch up on 24 hours of blocks in just under a minute. Any chance of getting the relative block count progress bar added back in as a configurable display option for the majority who liked it :) v0.6.1 perhaps. Keep up the good work! Title: Re: Version 0.6.0 released Post by: Cusipzzz on March 31, 2012, 12:58:06 PM kudos to all the developers!
Title: Re: Version 0.6.0 released Post by: rjk on March 31, 2012, 01:00:28 PM Initial network synchronization should be much faster (one or two hours on a typical machine instead of ten or more hours). Here's something I've been wondering for awhile now, but only now remember when I'm actually on the forums... Would it make sense to include a version for new adopters to download which includes an up-to-date (at time of release) blkindex file rather than a blank blkindex file so that they are nearly synced at installation? Title: Re: Version 0.6.0 released Post by: Pieter Wuille on March 31, 2012, 01:13:25 PM The new version downloads the blocks much faster now, but if that is still too slow you can download a nightly here: http://eu1.bitcoincharts.com/blockchain/ If you're going to suggest people to download the block chain database (bypassing all verification that bitcoin does when it fetches blocks from peers), it may be advisable to suggest using -checkblocks -checklevel=6 when starting up the first time. -checklevel is a new option that allows much more thorough verification. Otherwise, if somehow such a download (by bad luck or malicious intent) gets corrupted, and many people download their chain from that point, a particular block could put them all at once on a block chain fork. Title: Re: Version 0.6.0 released Post by: Mushoz on March 31, 2012, 02:44:41 PM False alarm - after 5 minutes or so it started going up. Same problem here. Getting stuck on 8 connections sometimes. uPnP doesn't seem to be working as well as it used to. Title: Re: Version 0.6.0 released Post by: Pieter Wuille on March 31, 2012, 03:12:51 PM Same problem here. Getting stuck on 8 connections sometimes. uPnP doesn't seem to be working as well as it used to. The reason is IRC being deprecated and defaulting to off. This will indeed mean less very fast incoming connections, but better connectivity for longing-living listening nodes (as others don't rely on recently IRC-advertized nodes, but on the addresses propagated through the network). Title: Re: Version 0.6.0 released Post by: Mushoz on March 31, 2012, 03:40:26 PM Same problem here. Getting stuck on 8 connections sometimes. uPnP doesn't seem to be working as well as it used to. The reason is IRC being deprecated and defaulting to off. This will indeed mean less very fast incoming connections, but better connectivity for longing-living listening nodes (as others don't rely on recently IRC-advertized nodes, but on the addresses propagated through the network). I see. Is it possible to get more than 8 connections on the initial download of the chain? The number of connections finally went past 8 when the download was very close to being finished. Maybe that had something to do with it? Title: Re: Version 0.6.0 released Post by: Pieter Wuille on March 31, 2012, 03:47:10 PM I see. Is it possible to get more than 8 connections on the initial download of the chain? The number of connections finally went past 8 when the download was very close to being finished. Maybe that had something to do with it? There's no point in that: downloading the chain will not go faster with more connections - one good one suffices. In fact, incoming connections have a much larger chance of wanting to request your block chain, slowing your sync down further. Title: Re: Version 0.6.0 released Post by: Portnoy on March 31, 2012, 03:48:26 PM Any chance of getting the relative block count progress bar added back in as a configurable display option for the majority who liked it :) v0.6.1 perhaps. +1 Title: Re: Version 0.6.0 released Post by: Mushoz on March 31, 2012, 04:47:03 PM I see. Is it possible to get more than 8 connections on the initial download of the chain? The number of connections finally went past 8 when the download was very close to being finished. Maybe that had something to do with it? There's no point in that: downloading the chain will not go faster with more connections - one good one suffices. In fact, incoming connections have a much larger chance of wanting to request your block chain, slowing your sync down further. I know it will only use 1 connection at a time for downloading, but with more connections there's a bigger chance you get connected to a fast peer, right? Title: Re: Version 0.6.0 released Post by: Pieter Wuille on March 31, 2012, 05:01:34 PM I know it will only use 1 connection at a time for downloading, but with more connections there's a bigger chance you get connected to a fast peer, right? No, it just uses whatever node is the first that tells you about a new block. That is almost always the first node you connect to. Title: Re: Version 0.6.0 released Post by: minorman on March 31, 2012, 06:45:40 PM Good work devs.
Bitcoin just keeps getting better... Title: Re: Version 0.6.0 released Post by: Daily Anarchist on March 31, 2012, 08:38:08 PM After issuing the sudo apt-get upgrade command I get the following message
Reading package lists... Done Building dependency tree Reading state information... Done The following packages have been kept back: bitcoin-qt 0 upgraded, 0 newly installed, 0 to remove and 1 not upgraded. Anybody know why it's not allowing me to upgrade as normal? Title: Re: Version 0.6.0 released Post by: LightRider on March 31, 2012, 11:09:32 PM Thanks again for all of your effort devs!
Title: Re: Version 0.6.0 released Post by: Daily Anarchist on March 31, 2012, 11:35:31 PM After issuing the sudo apt-get upgrade command I get the following message Reading package lists... Done Building dependency tree Reading state information... Done The following packages have been kept back: bitcoin-qt 0 upgraded, 0 newly installed, 0 to remove and 1 not upgraded. Anybody know why it's not allowing me to upgrade as normal? Problem solved! Just had to issue the install-qt command instead. There were some outdated dependencies holding it up, which I later autoremoved after replacing them with more recent dependencies during installation. Title: Re: Version 0.6.0 released Post by: Dusty on April 01, 2012, 07:20:28 PM Great news, thanks!
Title: Re: Version 0.6.0 released Post by: D.H. on April 02, 2012, 03:20:32 PM I installed v0.6.0 and apparently someone had finished the Swedish translation so I got it in Swedish, great! BUT, there seems to be a problem with the "&" that is included in some strings to indicate the shortcut key. For example, the "Send coins" tab looks like this in the Swedish translation: "amp; Skicka bitcoins". So, it seems that "&" has been used in the program instead of just "&".
It looks right on Transifex from what I can see, and in some places it works, for example the "New address" button on the "Receive coins tab" looks right. Title: Re: Version 0.6.0 released Post by: Diapolo on April 02, 2012, 04:02:19 PM I installed v0.6.0 and apparently someone had finished the Swedish translation so I got it in Swedish, great! BUT, there seems to be a problem with the "&" that is included in some strings to indicate the shortcut key. For example, the "Send coins" tab looks like this in the Swedish translation: "amp; Skicka bitcoins". So, it seems that "&" has been used in the program instead of just "&". It looks right on Transifex from what I can see, and in some places it works, for example the "New address" button on the "Receive coins tab" looks right. It had been fixed by a translator on Transifex, so the next time a language update is imported, that will be fixed :). Dia Title: Re: Version 0.6.0 released Post by: dooglus on April 03, 2012, 10:00:33 PM Any chance of getting the relative block count progress bar added back in as a configurable display option for the majority who liked it :) v0.6.1 perhaps. How about this? https://github.com/bitcoin/bitcoin/pull/1030 Show the number of blocks left to download, instead of a percentage. The reason the display was changed back to being absolute rather than relative was for new users. Suppose a new user installs the client, downloads half the blocks, sees it take hours to go from 0% to 50%, then quits and restarts. With the relative scheme, the display would start back to 0% again, and the user may well think that it was starting from the beginning, and give up on the whole idea. See the discussion here: https://github.com/bitcoin/bitcoin/issues/753 Title: Re: Version 0.6.0 released Post by: slothbag on April 03, 2012, 10:20:52 PM I understand why we went back to absolute and agree it needed to be done for the newbs.. What I was suggesting is an option to switch to relative in the options screen.
Title: Re: Version 0.6.0 released Post by: ThiagoCMC on April 04, 2012, 05:42:43 AM Hi!
Where is the TX-fees?!?! I just send 50.00 and 10.00 and for both transactions, the fee = 0.00 !! But, with the previous versions, the client automatically put a default fee there... I would like to pay the fees!! how to enable it again?! Thanks! Thiago Title: Re: Version 0.6.0 released Post by: Diapolo on April 04, 2012, 05:52:48 AM Hi! Where is the TX-fees?!?! I just send 50.00 and 10.00 and for both transactions, the fee = 0.00 !! But, with the previous versions, the client automatically put a default fee there... I would like to pay the fees!! how to enable it again?! Thanks! Thiago In the settings, you'll find a checkbox there! Or with a supplied flag, -paytxfee=. Dia Title: Re: Version 0.6.0 released Post by: ThiagoCMC on April 04, 2012, 06:04:35 AM Hi! Where is the TX-fees?!?! I just send 50.00 and 10.00 and for both transactions, the fee = 0.00 !! But, with the previous versions, the client automatically put a default fee there... I would like to pay the fees!! how to enable it again?! Thanks! Thiago In the settings, you'll find a checkbox there! Or with a supplied flag, -paytxfee=. Dia Mmmm... I don't use GUI... only bitcoind in Ubuntu Linux... So, I just need to add the option: paytxfee=?.?? at ~/.bitcoin/bitcoin.conf ? I would like to have it dynamic as before... Not a fixed fee. But how?! Title: Re: Version 0.6.0 released Post by: Pieter Wuille on April 04, 2012, 11:25:05 AM Hi! Where is the TX-fees?!?! I just send 50.00 and 10.00 and for both transactions, the fee = 0.00 !! But, with the previous versions, the client automatically put a default fee there... I would like to pay the fees!! how to enable it again?! Thanks! Thiago Fees have always been voluntary, except in case of sending very small outputs, or using too recent/too small inputs, as a means for spam protection on the network. The details of this policy have not been changed since 0.3.23. Title: Re: Version 0.6.0 released Post by: ThiagoCMC on April 04, 2012, 05:02:28 PM Well, the behavior in 0.6.0 is different of 0.5.3.
With 0.5.3 (using the same bitcoin.conf) I pay the fees automatically, now, I pay no fees... How to "restore" the old behavior?! Title: Re: Version 0.6.0 released Post by: Proofer on April 05, 2012, 02:41:54 AM Well, the behavior in 0.6.0 is different of 0.5.3. With 0.5.3 (using the same bitcoin.conf) I pay the fees automatically, now, I pay no fees... How to "restore" the old behavior?! Code: $ bitcoind help settxfee Title: Re: Version 0.6.0 released Post by: rjk on April 05, 2012, 02:43:58 AM Well, the behavior in 0.6.0 is different of 0.5.3. For those transactions, it is entirely possible that you wouldn't have paid a fee even with the old client. Fees are usually applied when the transaction is over the size limit, but they also may be waived if the coins have been sitting in your wallet for a sufficient period of time. I'd say don't worry about it.With 0.5.3 (using the same bitcoin.conf) I pay the fees automatically, now, I pay no fees... How to "restore" the old behavior?! Title: Re: Version 0.6.0 released Post by: Pieter Wuille on April 05, 2012, 03:02:54 AM Well, the behavior in 0.6.0 is different of 0.5.3. I guarantee you, nothing changed in the fee policy since 0.3.23. Whether a fee is required or not depends on the size and age of the input coins you're using. Title: Re: Version 0.6.0 released Post by: ThiagoCMC on April 05, 2012, 03:11:20 AM Okay! Thank you guys!! :)
Title: Re: Version 0.6.0 released Post by: jetmine on April 05, 2012, 04:12:25 PM Gavin Andresen,
Horrible work. Absolutely NOT impressed. You have >100 bugs open and yet decide to release on 30-march and FORCE the update for 1-april. Just two days later. This release is unstable as hell. I get crashes. I get unexpected exits. And I get failures to restart. Crashes: ----------- ************************ EXCEPTION: 22DbRunRecoveryException DbEnv::txn_checkpoint: DB_RUNRECOVERY: Fatal error, run database recovery bitcoin in ThreadMessageHandler() Unexpected exits: -------------------- 04/05/12 11:38:48 sending: inv (73 bytes) 04/05/12 11:38:49 received: addr (31 bytes) connection timeout 04/05/12 11:38:49 received: addr (31 bytes) trying connection 200.74.242.32:8333 lastseen=-2.3hrs 04/05/12 11:38:50 received: addr (31 bytes) 04/05/12 11:38 --> Result code: 139 Failures to restart: -------------------- Loading addresses... dbenv.open strLogDir=/home/XXX/.bitcoin/database strErrorFile=/home/XXX/.bitcoin/db.log ************************ EXCEPTION: NSt8ios_base7failureE CDataStream::read() : end of data bitcoin in AppInit() --> Result code: 134 What about good Q&A procedures? What about testing stuff before release ("compiles fine on Win7" is NOT testing)? What about establishing a stable status quo BEFORE forcing it on people, like for a month/year? I get 80% orphans with the old version, and I get 95% downtime with the new version. Great choices, thanks a lot for that. Title: Re: Version 0.6.0 released Post by: rjk on April 05, 2012, 04:15:53 PM What about good Q&A procedures? What about testing stuff before release ("compiles fine on Win7" is NOT testing)? What about establishing a stable status quo BEFORE forcing it on people, like for a month/year? The devs have been begging for beta testers for a while now. I suggest that you liaise with them to set something up, instead of bashing them without knowing what's up. Perhaps you could join the #bitcoin-dev IRC channel on freenode, and hit up the mailing list, and give them REPRODUCIBLE issues to test and fix.Calm down. Title: Re: Version 0.6.0 released Post by: DeathAndTaxes on April 05, 2012, 04:20:51 PM This release is unstable as hell. I get crashes. I get unexpected exits. And I get failures to restart. Good thing you reported this in RC1 to RC5 so it is already fixed..... er wait you didn't? Title: Re: Version 0.6.0 released Post by: Boussac on April 05, 2012, 04:49:17 PM I wonder what is the criteria to become "whitelisted" ? (one must have been angry for the past six months ?)
@gavin and team: thanks and keep up the good work, you guys are outstanding! Title: Re: Version 0.6.0 released Post by: Gavin Andresen on April 05, 2012, 05:58:32 PM This release is unstable as hell. I get crashes. I get unexpected exits. And I get failures to restart. All righty... you're the first person I've heard that from.Please file bug reports on the github issue tracker. Include enough information so we can reproduce the problem (what platform? what seems to cause the problem? etc) and it might get fixed. I say "might" because if you're serving up a big mining pool running bitcoind on Windows... well, none of the current core developers work on Windows. Title: Re: Version 0.6.0 released Post by: jetmine on April 05, 2012, 07:14:15 PM Please file bug reports on the github issue tracker. Include enough information so we can reproduce the problem (what platform? what seems to cause the problem? etc) and it might get fixed. Github doesnt seem to accept new accounts: "There were problems creating your account." It gives no further details, so I dont know how to get on there. The platform is CentOS 5.6 x64. The cause of the problem is unknown. It started shortly after 2am CET this morning, and is probably related to whatever has been received around that time (because it worked for almost a day before). I have posted everything that I have, which is (-debug -printtoconsole). There was another piece of information which I didnt save unfortunately. It a crash report after the first crash, and it contained a call stack dump. The initial cause for the crash was a "free" of invalid memory. I have another similar box that also crashed tonight, but it did not dump the callstack and it did have less problems to start again. Therefore the other problems are probably caused by data that has been saved during the initial problematic event, and in fact the first box only came up again after deleting all cached data in .bitcoin/ and in particular the addr.dat file. So there must be at least three problems, one is that an invalid memory free can be triggered, another one is that the file parsers can bail out fatally when encountering corrupt file content, and the third problem is that normal operation can cause files be filled with such corrupt content. Let me know what else you need to know from me. Title: Re: Version 0.6.0 released Post by: pandemic on April 05, 2012, 09:51:03 PM Any issues upgrading or can I just install and open w/o loss?
Title: Re: Version 0.6.0 released Post by: DeathAndTaxes on April 05, 2012, 09:51:54 PM Any issues upgrading or can I just install and open w/o loss? You can just install but always (on this ver or any verion) make a backup of wallet.dat before upgrading. Title: Re: Version 0.6.0 released Post by: Red Emerald on April 05, 2012, 09:56:36 PM The blockchain download speed is soo much faster now!
<3 Title: Re: Version 0.6.0 released Post by: SkRRJyTC on April 05, 2012, 10:40:09 PM The devs have been begging for beta testers for a while now. How would I begin to get involved with this? Title: Re: Version 0.6.0 released Post by: ThiagoCMC on April 05, 2012, 11:27:17 PM Any issues upgrading or can I just install and open w/o loss? What I've done: 1- Stop your Bitcoin <0.5.3; 2- Copy your wallet; 3- remove (or move it away) the "old" blockchain (i.e. mv ~/.bitcoin/ ~/_bitcoin-0.5.3-backup); 4- create a empty dir (i.e. mkdir ~/.bitcoin/); 5- copy the wallet.dat from the backup to ~/.bitcoin; 6- start Bitcoin 0.6.0. Title: Re: Version 0.6.0 released Post by: dlb76 on April 06, 2012, 01:17:28 AM updated to 0.6.0.6-beta
Thanks for QR codes! Title: Re: Version 0.6.0 released Post by: GideonGono on April 06, 2012, 05:43:25 AM Thanks to all the contributors to this release!!
Title: Re: Version 0.6.0 released Post by: jetmine on April 06, 2012, 09:24:56 PM Please file bug reports on the github issue tracker. Include enough information so we can reproduce the problem (what platform? what seems to cause the problem? etc) and it might get fixed. Another crash happened today, same platform etc. This time it ran with -printtoconsole (no -debug). The scrollback buffer is filled with hundreds of Code: received getdata for: tx 7680e65403ab1b258887 .. then: Code: received getdata for: tx 7680e65403ab1b258887 ... Code: Added 1 addresses from 76.117.220.29: 3415 tried, 8128 new .. and now I get literally hundreds of this: Code: ERROR: CBlock::ReadFromDisk() : OpenBlockFile failed .. after which it continues like this: Code: ERROR: CBlock::ReadFromDisk() : OpenBlockFile failed .. and here bitcoind is dead. No more messages, nor a stackdump either. Resultcode is 134. I got the same behaviour on two different boxes. The earliest abnormal behaviour I could find in the scrollback buffer is transaction 7680e65403ab1b258887 (dont know if this identifier is meaningful after bitcoind has shutdown?). Then there is that sole "OpenBlockFile failed", which shortly extends into hundreds of the same error which take down all dependent operations very quickly. This happened around 7pm CET this afternoon. If necessary, I think I have full -debug -printtoconsole logs on the other box (not limited by scrollback buffer size). Let me know if you need them and I will recover them for you. Edit: added code tags Title: Re: Version 0.6.0 released Post by: rjk on April 06, 2012, 09:38:08 PM Look to me like it could be either disk corruption or someone flooding your node with an invalid tx. I think you mentioned that you have already blown away the block database once, but if you haven't that would be something to try.
Title: Re: Version 0.6.0 released Post by: jetmine on April 06, 2012, 09:38:36 PM Another good one:
... Code: version message: version 40000, blocks=174571 The returncode is 139 at this point. Note how at one point is sais received block 00000000000008506c9e, then it sais error, already have 00000000000008506c9e. Then shortly after it sais ask for block 00000000000008506c9e and shortly after it just segfaults. Edit: added code tags Title: Re: Version 0.6.0 released Post by: paraipan on April 06, 2012, 09:42:54 PM interesting...
Quote askfor tx 263d4ec0b5712e6e1b47 1333717043000000 Quote askfor block 00000000000008506c9e 1333747883000000 Title: Re: Version 0.6.0 released Post by: jetmine on April 06, 2012, 09:44:26 PM Look to me like it could be either disk corruption or someone flooding your node with an invalid tx. I think you mentioned that you have already blown away the block database once, but if you haven't that would be something to try. Disk corruption is not likely. The two boxes have different hardware, different blockdevs, even different filesystems. Wipe database, yes I did this already. About your other suspicion: flood with invalid tx -> crash == DoS attack vector. The goold old 3.xx branch was rock stable, and now we are FORCED to use software which has less than a week of testing (and many obvious problems already visible). Title: Re: Version 0.6.0 released Post by: rjk on April 06, 2012, 09:49:04 PM About your other suspicion: flood with invalid tx -> crash == DoS attack vector. Some similar tx flood DoS attack vectors were recently closed, but it is entirely possible that more remain, or that the patch has opened a different vector. It will be interesting to find out the cause of this crashing.BTW, I believe that Luke-jr has posted backports of 0.5.x, if that interests you. Title: Re: Version 0.6.0 released Post by: jetmine on April 06, 2012, 10:00:00 PM What about this one:
Bitcoin version 0.6.0.99-beta Default data directory /home/USERNAME/.bitcoin Loading addresses... dbenv.open strLogDir=/home/USERNAME/.bitcoin/database strErrorFile=/home/USERNAME/.bitcoin/db.log ************************ EXCEPTION: NSt8ios_base7failureE CDataStream::read() : end of data bitcoin in AppInit() Result code is 134 and it repeats like this until I delete addr.adt Apparently addr.dat can also be filled with "poisonous data" from remote. Good way to take offline your competitors. Title: Re: Version 0.6.0 released Post by: Gavin Andresen on April 07, 2012, 12:02:44 AM Is anybody else seeing anything like what jetmine is seeing? Anybody else running CentOS 5.6? Did you compile from source or are you using the binaries we compiled?
My 0.6 nodes running on Ubuntu 11 have been rock solid. The "CBlock::ReadFromDisk() : OpenBlockFile failed" is very odd, that should never happen. You aren't running with a -datadir on a network drive or something are you? RE: filling addr.dat: that is one of the denial-of-service attacks fixed by the 0.6 release. Title: Re: Version 0.6.0 released Post by: DeathAndTaxes on April 07, 2012, 02:49:52 AM Nope but not using Centos. Have 3 copies running 1 on Win7 and 2 on Debian. They support a farm running p2pool. Haven't had a single crash on any instance.
Title: Re: Version 0.6.0 released Post by: jgarzik on April 07, 2012, 08:28:58 AM run it in gdb, and use the "bt" command when a segfault occurs. Make sure to enable thread tracing, if that is not default / built in.
Title: Re: Version 0.6.0 released Post by: jetmine on April 07, 2012, 11:58:51 AM Did you compile from source or are you using the binaries we compiled? The "CBlock::ReadFromDisk() : OpenBlockFile failed" is very odd, that should never happen. You aren't running with a -datadir on a network drive or something are you? Compiled from source. The datadir is on the local disk (LVM volume on the same physical disk as the OS). run it in gdb, and use the "bt" command when a segfault occurs. Make sure to enable thread tracing, if that is not default / built in. Thanks, I will do that. One doubt though. How can I enable thread tracing, or check whether it is enabled by default? Google tells me to use ptrace instead of gdb for thread tracing, but I suspect this is not what you want me to do. Title: Re: Version 0.6.0 released Post by: BlackEye on April 07, 2012, 03:33:46 PM I've been running the released 0.6.0.6-beta Windows version for a few days without issue. This morning I see that it had crashed.
Code: AcceptToMemoryPoolUnchecked(): size 165 Title: Re: Version 0.6.0 released Post by: jetmine on April 08, 2012, 09:56:35 AM RE: filling addr.dat: that is one of the denial-of-service attacks fixed by the 0.6 release. I am running both 0.6.0 release and the master from Apr-03 (which is shortly afterwards). Yet I get these addr.dat errors. There must be another bug to fix in addr.dat Today I got two other crashes, one of them in gdb. I will post it in a separate post, so that my point about addr.dat doesnt get drowned. Title: Re: Version 0.6.0 released Post by: jetmine on April 08, 2012, 10:05:27 AM run it in gdb, and use the "bt" command when a segfault occurs. Make sure to enable thread tracing, if that is not default / built in. Ok, so this is a crash with v0.6.0 release at around 6am today. I still dont know how to enable thread tracing, so I hope it was just enable by default (?). The backbuffer of -debug -printtoconsole looks quite normal: Code: Added 1 addresses from 91.154.226.179: 3470 tried, 11409 new Now comes a dump that was made automatically. It looks similar to what I mentioned that had happened already once to me: Code: ======= Backtrace: ========= When I use the bt command, I get this: Code: #0 0x000000000082ce95 in raise () I have to leave for a few hours, but will keep the debugger open. If there are any commands that can extract more useful information from it, please post here and I will run them on it tonight. Edit: Added code tags (good idea). BTW, the box is still up, for if anyone wants to see the output of another gdb command, memory dump, etc. Title: Re: Version 0.6.0 released Post by: Red Emerald on April 08, 2012, 05:50:40 PM jetmine. You should really put those giant posts inside code blocks
Code: Like this Title: Re: Version 0.6.0 released Post by: Gavin Andresen on April 08, 2012, 07:24:59 PM When I use the bt command, I get this: Thanks, that should be very helpful. I opened issue #1065 https://github.com/bitcoin/bitcoin/issues/1065.... |