Armory - Discussion Thread

[Wotan777]

...

Ugh, I just realized I bet we broke the 12.04 offline bundle with this release.  I bet no one tested that... we really need to widen the testing net better.  In the future, we might offer quadruple bounties for the first bug found in each of a bunch of categories.  That would include testing the offline bundles, and lots of corner case features...

I might need to update the offline bundles and re-release 0.93...

Armory-0.93 Win7 64 bit version (Version 0.93-beta-84a5897ec2) also has a problem:
Today it crashed and rebuilt Armory database 3 times.

[1714043545]

1714043545

[1714043545]

1714043545

[1714043545]

1714043545

[1714043545]

1714043545

[bitcool]

The latest version has memory leaks, badly. It consume 15.8 GB of my 16 GB Windows memory, every time.

[goatpig]

The latest version has memory leaks, badly. It consume 15.8 GB of my 16 GB Windows memory, every time.

What's Armory's RAM usage? I think you are confusing RAM used by the process and in use memory pages. The OS won't shy away from filling that up when a lot of access is performed on mmap'ed files, which will happen a lot during a DB scan. That doesn't mean the RAM is in use, it means the OS is chugging unused RAM to shuffle file data in and out of it. If a process was to ask for more RAM it would get priority over this memory.

The OS has no reason to empty this memory once Armory is done scanning. For all it knows, Armory just asked for intensive access to this data, so the OS expects it will again. Unless other processes start asking for more RAM or Armory is closed, the OS will keep these unused pages filled with the DB's mmap'ed files.

[bitcool]

The latest version has memory leaks, badly. It consume 15.8 GB of my 16 GB Windows memory, every time.

What's Armory's RAM usage? I think you are confusing RAM used by the process and in use memory pages. The OS won't shy away from filling that up when a lot of access is performed on mmap'ed files, which will happen a lot during a DB scan. That doesn't mean the RAM is in use, it means the OS is chugging unused RAM to shuffle file data in and out of it. If a process was to ask for more RAM it would get priority over this memory.

The OS has no reason to empty this memory once Armory is done scanning. For all it knows, Armory just asked for intensive access to this data, so the OS expects it will again. Unless other processes start asking for more RAM or Armory is closed, the OS will keep these unused pages filled with the DB's mmap'ed files.

I think you are correct. It's probably not a memory leak, ArmoryQt itself only uses 300M, on top of bitcoind's 500M.  However it's still scary to see so much memory being mmap'd, rarely have I seen this before.

Thanks for pointing that out.

[goatpig]

The latest version has memory leaks, badly. It consume 15.8 GB of my 16 GB Windows memory, every time.

What's Armory's RAM usage? I think you are confusing RAM used by the process and in use memory pages. The OS won't shy away from filling that up when a lot of access is performed on mmap'ed files, which will happen a lot during a DB scan. That doesn't mean the RAM is in use, it means the OS is chugging unused RAM to shuffle file data in and out of it. If a process was to ask for more RAM it would get priority over this memory.

The OS has no reason to empty this memory once Armory is done scanning. For all it knows, Armory just asked for intensive access to this data, so the OS expects it will again. Unless other processes start asking for more RAM or Armory is closed, the OS will keep these unused pages filled with the DB's mmap'ed files.

I think you are correct. It's probably not a memory leak, ArmoryQt itself only uses 300M, on top of bitcoind's 500M.  However it's still scary to see so much memory being mmap'd, rarely have I seen this before.

Thanks for pointing that out.

If you restart Armory it shouldn't be mmap'ing as much memory. Next version will be a lot tamer on that front, and I also may just close and reopen the DB just to flush out the mmap data after the scan is over. For now you'll have to deal with the scary numbers =P

[goatpig]

why is my blockchain stuck?

still with Armory 0.92.3 and Bitcoin Core 0.9.3 in Ubuntu 12.04 LTS 32 bit.

Care to elaborate?

[cypherdoc]

why is my blockchain stuck?

still with Armory 0.92.3 and Bitcoin Core 0.9.3 in Ubuntu 12.04 LTS 32 bit.

Care to elaborate?

nvm, fixed.

when is the 0.93 update going to work with my OS?

[goatpig]

nvm, fixed.

when is the 0.93 update going to work with my OS?

You mean the 32bit part or the 12.04 part?

We got some fixes incoming soon that will get Armory going again on 12.04 among other things. For x86 support you'll have to wait for the next subversion.

[zveda2000]

Hi,

I keep getting the error "Armory scanning interrupted unexpectedly". Armory then asks me to check the box to re-scan and rebuild the database on next time Armory starts. If I do not check the box I get prompted with the same error on restart. This happened two times already in the last few days.

I'm running the latest armory on 64-bit linux with 16gb ram. My internet connection is fairly slow though, especially upload speed.

[goatpig]

Hi,

I keep getting the error "Armory scanning interrupted unexpectedly". Armory then asks me to check the box to re-scan and rebuild the database on next time Armory starts. If I do not check the box I get prompted with the same error on restart. This happened two times already in the last few days.

I'm running the latest armory on 64-bit linux with 16gb ram. My internet connection is fairly slow though, especially upload speed.

https://bitcointalk.org/index.php?topic=970983.0

[zveda2000]

I think I'm all good now after updating to fixes branch. Thanks!

[goatpig]

I think I'm all good now after updating to fixes branch. Thanks!

Post that in the other thread so I can keep track =)

[MoreBloodWine]

Ok, so I got cor 0.10 running, went to dl Armory 0.93 which afaik is supposed to be 0.10 compatable.

Anyway, when I run it I get some windows version error tellin me to dl a 32 or 64 bit compatable version. So what am I doing wrong here, previous version of armory ran fine but 0.93 seems fubar'd to me.

Ty.

[doug_armory]

Ok, so I got cor 0.10 running, went to dl Armory 0.93 which afaik is supposed to be 0.10 compatable.

Anyway, when I run it I get some windows version error tellin me to dl a 32 or 64 bit compatable version. So what am I doing wrong here, previous version of armory ran fine but 0.93 seems fubar'd to me.

Ty.

Do you have a 32-bit version of Windows? Due to an oversight, Armory 0.93 can't be installed on them. If using for offline-only, you'll need to roll back to 0.92.3 for now. If using for online, you'll have to wait and see if service is restored. (Despite the oversight, running online Armory on a 32-bit system really can't be done just yet anyway.)

[Rampion]

Rampion, can you file a support ticket and include both your logs (armorylog.txt and armorycpplog.txt) and the stack trace that comes up when the code crashes?

In the meantime, can you try something for me? Can you create a directory called "ArmorySatoshiTest" and then run Armory from the command line? Use the following steps, with the assumption that the directory is on the desktop.

1)Open the terminal.
2)cd *Armory location*/Armory.app/Contents/MacOS
3)./Armory --satoshi-datadir="/Users/*your username*/Desktop/ArmorySatoshiTest"

This will cause the blockchain to be downloaded again. Sorry! We definitely need this step to troubleshoot what's going on, unfortunately.

Thanks!

Sorry for the late reply - I've been away from my computer in the last two weeks.

I already sent some logs on 20/2 - I have since then reverted to 0.92.3 (and still using BC 0.10.0) and for now everything is working OK.

[Muhammed Zakir]

Is there a way to convert normal raw TX to Armory compatible TX? I found a simple script which does it. So would be helpful if there is a script for multisig TX. Thanks! Any opinions/suggestions are welcome!

   -MZ

[etotheipi]

Just pushed goatpig's fix into 0.93.0.70.  Yeah weird version number. This will be 0.93.1 after we've confirmed it does what it's supposed to.

As usual, get it from the secure downloader within Armory (help->update software).

  Armory 0.93.0.70-testing for Windows XP, Vista, 7, 8+ (64-bit)
  Armory 0.93.0.70-testing for MacOSX 10.7+ (64bit)
  Armory 0.93.0.70-testing for Ubuntu 12.04+ (32bit)
  Armory 0.93.0.70-testing for Ubuntu 12.04+ (64bit)
  Armory 0.93.0.70-testing for RaspberryPi  (armhf)

  Armory 0.93.0.70-testing Offline Bundle for Ubuntu 12.04 exact (32bit)
  Armory 0.93.0.70-testing Offline Bundle for Ubuntu 12.04 exact (64bit)
  Armory 0.93.0.70-testing Offline Bundle for RaspberryPi  (armhf)

  Armory 0.93.0.70-testing: Signed hashes of all installers

[MoreBloodWine]

Ok, so I got cor 0.10 running, went to dl Armory 0.93 which afaik is supposed to be 0.10 compatable.

Anyway, when I run it I get some windows version error tellin me to dl a 32 or 64 bit compatable version. So what am I doing wrong here, previous version of armory ran fine but 0.93 seems fubar'd to me.

Ty.

Do you have a 32-bit version of Windows? Due to an oversight, Armory 0.93 can't be installed on them. If using for offline-only, you'll need to roll back to 0.92.3 for now. If using for online, you'll have to wait and see if service is restored. (Despite the oversight, running online Armory on a 32-bit system really can't be done just yet anyway.)

Of course it's 32 bit, anyway, hope the oversight is fixed.

That being said, can I use my own rpc user / pass with armory or does it have to be those supplied when installing armory ?

[KS]

Trying to run core 0.10 and armory 0.93 on win7 x64. Neither can access the default db/blocks path.

For core, I think it's an issue with accented characters in the path (since 0.9, 0.8.6 was fine), so I changed the default bitcoin blockchain folder. Armory 0.92.3 was running fine with the changed --satoshi-datadir path  and was at least starting fine and showing me the wallets and transactions with core 0.10.

Upgraded to Armory 0.93 (disabled core auto start), first run with Armory 0.93: crashed when using the --satoshi-datadir argument. Start again without the --satoshi-datadir argument, seems to load fine, asks me if i wanted to delete the old db version from Armory 0.92.3, OK so far, then error message, cannot access the db in the user\appdata\roaming etc folder.

I did specify the correct paths in Armory for core and db but that didn't work (same issue with core 0.9.3 and armory 0.92.3). Is this related to core or Armory? (accented characters issue?))

In the end, I also changed the db path for Armory and am now running it with both --satoshi-datadir and --dbdir arguments. It's launched OK, sync'd OK, db re-built OK, scanning tx.

I wish there were a way to keep the settings within Armory and not in the program's link. Is this sth that can be fixed? (.ini file?)

[btchris]

I wish there were a way to keep the settings within Armory and not in the program's link. Is this sth that can be fixed? (.ini file?)

I'm with you, but then where would the .ini file be (and what if it were located in a path with non-ASCII characters)?

The "right" way to do this is of course to store settings in the Windows registry, however the Armory devs have to support multiple OS's, and it's understandable that they don't want to have to maintain different code paths for every OS.

On the plus side, an Armory dev did say that they plan to fix the non-ASCII-path issue over here: https://bitcointalk.org/index.php?topic=984101.msg10727929#msg10727929

Another way to go about this would be to migrate all of the settings storage to QSettings. They already use Qt, and QSettings abstracts away the "right" thing and does it pretty well (it does use the Windows registry for example). Of course, every minute they spend fixing these types of issues is a minute they don't spend implementing something more important....