Topic: Armory - Discussion Thread - page 61. (Read 521829 times)

We have tried to respond to as many support emails as we can.  But we are a dev team, not a customer support team, so we can't get to everything.  Since we hadn't isolated the cause of your issue yet, the majority of emails we received from users with your symptoms we could not respond intelligently. 

Like I said, we want to respond to everyone, but we're an open-source project that makes no money (yet) and we have no customer support staff.  We respond to the emails that have the easiest response.  And we certainly take note of of the problems we see. 

On that note, I think we found a solution to this.  We may be able to get it into 0.91 after all.  We're excited that many of users reporting these symptoms may be resolved with this fix (hopefully all of them!)

More likely support emails get zero reply.
At least that has been my experience for more than a week.

I am trying to run Armory with the latest 64-bit version of Bitcoin Core.  Is this not possible?  I have pointed Armory to the installation directory of Bitcoin Core, but it doesn't seem to recognize it.

Nevermind, Just had to point Armory to C:\Program Files\Bitcoin\daemon

What error are you referring to? There was a leveldb bug that got fixed with the latest testing build (the one posted tonight) on Windows. Your post infers this bug occurs on Linux too, where it is built stock from google's source, so I'm bit curious now.


Most of these bugs are one and the same. Once we come to a fix the only thing we can do while we're working towards a release is to tell the users to wait for it. I guess it's better having someone doing that, but that's on our free time, so our presence can grow thin during busy times, like pre-release for these past 2 weeks. Also a lot of these users resort to sending us an email to our support channel, and those have a higher rate of resolution than what you can see here.

Your assumption on Armory's stability is a little off however. What you are experiencing is obviously a wallet issue as it carries from OS to OS. Changing OS only changes how resources are delivered to Armory. While we had problems with that for a while, these were eventually fixed, and the users left with an unusable Armory are those falling in your category, with a wallet pointing at a lot of large transaction. It took us some time to isolate this issue but that we're on it, it won't last that long.

For your information, I loaded Zoella's wallet in about 15sec. Coming from someone that simply can't get Armory to work, I'd say it's a step forward. Still implementing some fixes on that front as we speak. Back to the code now.

Btw, I just posted a new testing version and forgot to link to it from here:

0.90.99.3-testing is posted for both Windows and Ubuntu 12.04+

Unfortunately, we upgraded the makefiles, etc, and OSX building broke.  We don't think it will be tough to fix, but gotta fight through it.

---

@rocks -- we do take every report seriously, but we can't magically fix things just because a report was made.  There's been a lot of reports of the BDM hanging that we were only recently been able to identify.  Until we isolate the issue, we can't do anything about it.

In your case, there is obviously a pattern, because we test it regularly on new installs, and many users install on clean installs without issue.  (the exception is WinXP 64, which has always had problems).

--Are you using a paritcular wallet on each system?
--Are you using a particular bitcoin.conf?
--Are these all on the same computer?  What are its specs?
--Are you installing in non-standard locations?  
--Do you use command-line flags?
--Do you have non-english characters in your path names or use them in the interface?
--Do you run bitcoind/bitcoin-qt manually, or auto-managed?

As for the leveldb-error... I'm not sure exactly what that is.  Is it an error that there's a problem with the databases?  Or leveldb itself appears to be missing?  If leveldb is there but reports problematic files, try running with --rebuild.  

Again, I'm not here to discredit you or say anything is unimportant.  But clearly something is awry, as we've had plenty of users report success with every one of those environments (except WinXP64).  It would be great to know what is different, if anything.

"Unusaby crash-prone" is an understatement, at this point I've tried the following setups over the past 2-4 weeks to try and get Armory working but still with zero luck. All systems are spec'ed well enough.  

0.88, WinXP 64, existing system       - bitcoind stable, Armory crashes on sync
0.88, WinXP 64, clean install            - bitcoind stable, Armory crashes on sync
0.90, Win7, existing system             - bitcoind stable, Armory crashes on sync
0.90, Win7, clean install                  - bitcoind stable, Armory crashes on sync
0.90, Ubuntu 12.04 LTS, clean install - bitcoind stable, Armory crashes after ~10ish minutes
0.91 test, Win7, clean install            - bitcoind stable, Armory crashes on startup (leveldb missing error)
0.91 test, Ubuntu 12.04, clean install, compiled from github - bitcoind stable, Armory crashes on startup (leveldb missing error)

I've tried the above systems under multiple setups, first with .bitcoin and .armory located on a remote ZFS samba server (which I use for all home files) and then switched to default locations on the local host drive. I've also tried using my existing blockchain and letting bitcoind re-download the blockchain from scratch at least 3 different times. I am not an "average user".

The failure to run on a fresh Ubuntu 12.04 LTS system both current version and github latest beta to me just speaks volumes. It's a shame because I'd like the combined features of deterministic wallets together with multiple wallets and Armory is the only option that does that currently does this, but Armory just does not run. I've asked in other threads what to do and what error logs to post, but the developers sounded offended and didn't offer any real suggestions.

I think kentt is right and many try, fail and leave without posting. Since I've started to pay attention to the Armory forum about a month ago I've also seen several threads on crashing for which little help was offered and no resolution provided. I've been unusually determined to try many different configurations to get it to work and at this point am probably close to 24 hours of screen time, the vast majority of users I'm betting abandon before that.

If the devs are serious about making Armory a real wallet then they need to take every single post on Armory not working or crashing seriously and bring a resolution to each and every posted issue, otherwise the bugs pile up until Armory is not functional for most users in most situations.

Sorry if this became a bit of a rant, but I've spent a silly amount of time on this.

FYI, got a response saying I needed to submit a trouble ticket. Might want to check your support email account.

We're aware of this and this has been fixed in 0.91. As far the communication protocol goes, 0.90 is compatible with core 0.9.0 regardless

Sent, let me know.

Okay great.  I hadn't realized that you had accepted this as the issue.  It sounds like you've got a handle on the issue.
Thanks.

Absolutely.  Without it we'll have to manually hack together a wallet that exhibits the problem.  It would be much easier if we had one we already know causes problems. 

FYI we're stirring up our support email channel, so I'm not sure if a simple email with attachment will work.  Please try it and we'll find another way to get it if doesn't work (just support at bitcoinarmory dot com).

Just one tiny thing: Armory doesn't seem to find the daemon after installing core 0.9.0. Had to manually point it at Bitcoin\daemon.

Do you still need a watching only wallet? I've transferred everything out of mine and plan to retire it. Let me know how to send one if you need it.

The point is that these transactions have identified an inefficiency in Armory that will have to be fixed.  However, I believe that most users do not have such transactions, and it is only now that we've been able to identify that as the cause for a lot of these reports.  Because no one on our team has a wallet with such transactions in it.  Those types of transactions simply do not exist in many use cases, and thus many people will not experience any such problems.

Please do not misinterpret my statement as blaming it on the user, or saying it's unimportant.  I'm simply identifying that you have fallen into a gap that many users do not, and thus will not have those problems.  Now that we've identified it, we'll try to get a fix into 0.91.1.  Fixing it will require a bit more work than we can put in for 0.91.

No you don't. It's preferable that you do but 0.90 is compatible with bitcoin core 0.9.0

These are the typical transaction that we identified bring Armory to a crawl, we're building a test case to identify possible inefficiencies. Worst case scenario we'll have to overhaul that part.

Heavy testing?  Can you take 10 transactions like this? http://blockchain.info/tx/e3aea19a966d103c4520530f6a8f9486b03d4cbe46deeeb56c4e72730fe83f2f

I couldn't on any rig.

Same here ... btw, do I need to update something now, when qt 0.9 is out?

I have 0.90 installed for ages (it feels). Never a single problem, no crash, no rescan, nothing. Works like a charm here on Debian.

Ente

Ahh, gotcha. An intermediate wallet would make sense.