Author Topic: October 5 Test Network  (Read 56692 times)

0 Members and 1 Guest are viewing this topic.

jakub

  • Guest
Re: October 5 Test Network
« Reply #300 on: October 07, 2015, 06:58:26 pm »
bitshares-argentina  can you please let me know if you have changed your software in any way because it appears your node is consistently producing blocks even though you are not in the set of active witnesses.

BM, does this strange event have an explanation now or is it still unclear what happened?

Offline rnglab

  • Full Member
  • ***
  • Posts: 171
    • View Profile
  • BitShares: rnglab
Re: October 5 Test Network
« Reply #301 on: October 07, 2015, 06:58:34 pm »
bitshares-argentina  can you please let me know if you have changed your software in any way because it appears your node is consistently producing blocks even though you are not in the set of active witnesses.

Everyone else, please update your witness to the latest master before block 58500  (about 4 hours).

Hi dan, I didn't change anything. Neither played with uncommon settings as far as I remember . 

Running from master branch with last commit db84a492b9a2cf290f92a3987b59c11808138d56

How can I help debugging?









Offline Thom

Re: October 5 Test Network
« Reply #302 on: October 07, 2015, 06:59:46 pm »
The accuracy of each individual computers clock can be off by up to 300 ms.   It just means that your local clock is ahead of the other peer by 23 ms plus the network latency in fetching the block.

So there is a margin of error of 300ms in latency numbers?

That makes it pretty tough to get an accurate picture. I thought ntp would take care of that, not so?
Injustice anywhere is a threat to justice everywhere - MLK |  Verbaltech2 Witness Reports: https://bitsharestalk.org/index.php/topic,23902.0.html

Offline rnglab

  • Full Member
  • ***
  • Posts: 171
    • View Profile
  • BitShares: rnglab
Re: October 5 Test Network
« Reply #303 on: October 07, 2015, 07:03:11 pm »
I am getting this sometimes

Code: [Select]
3033114ms th_a       db_block.cpp:191              _push_block          ] Failed to push new block:
10 assert_exception: Assert Exception
next_block.witness == scheduled_witness: Witness produced block at wrong time
    {"block witness":"1.6.37","scheduled":"1.6.24","slot_num":1}
    th_a  db_block.cpp:646 validate_block_header

    {"next_block.block_num()":58718}
    th_a  db_block.cpp:511 _apply_block
3033116ms th_a       application.cpp:425           handle_block         ] Error when pushing block:
10 assert_exception: Assert Exception
next_block.witness == scheduled_witness: Witness produced block at wrong time
    {"block witness":"1.6.37","scheduled":"1.6.24","slot_num":1}
    th_a  db_block.cpp:646 validate_block_header

    {"next_block.block_num()":58718}
    th_a  db_block.cpp:511 _apply_block

    {"new_block":{"previous":"0000e55de39c07283d1779c89018c18206bfa9f2","timestamp":"2015-10-07T17:50:33","witness":"1.6.37","transaction_merkle_root":"0000000000000000000000000000000000000000","extensions":[],"witness_signature":"1f19b73fc82aa7519e28f34b1f8e10f63da27365c8e5699fd9f77c0ee1f4c4b02841560af821e3ec8e71f72264934140c7fb2cdd78e0f91920edc7b8cba9c33cfb","transactions":[]}}
    th_a  db_block.cpp:197 _push_block

I see the same. Perhaps something wrong wrong 1.6.37.

updated to latest master again (jtm1).

This particular witness has been producing blocks at the wrong time this whole time and fortunately helped us identify a bug that we fixed with a hard fork that took effect and is now generating these error messages.

@bytemaster let me know if I should update 1.6.37 now or if you need  it for debugging
« Last Edit: October 07, 2015, 07:11:44 pm by rnglab »

Offline bytemaster

Re: October 5 Test Network
« Reply #304 on: October 07, 2015, 07:12:37 pm »
This particular witness has been producing blocks at the wrong time this whole time and fortunately helped us identify a bug that we fixed with a hard fork that took effect and is now generating these error messages.

[member=5]bytemaster[/member] let me know if I should update 1.6.37 now or if you need  it for debugging

You should upgrade to the latest master and reindex. 
For the latest updates checkout my blog: http://bytemaster.bitshares.org
Anything said on these forums does not constitute an intent to create a legal obligation or contract between myself and anyone else.   These are merely my opinions and I reserve the right to change them at any time.

Offline betax

  • Hero Member
  • *****
  • Posts: 808
    • View Profile
Re: October 5 Test Network
« Reply #305 on: October 07, 2015, 07:16:50 pm »
BM, I think there are now 3 witnesses waiting to be voted in:

bitshares-argentina
triox-delegate
pmc

Voted.   Any others?

betaxtrade (please)
https://metaexchange.info | Bitcoin<->Altcoin exchange | Instant | Safe | Low spreads

Offline bytemaster

Re: October 5 Test Network
« Reply #306 on: October 07, 2015, 07:21:13 pm »
The accuracy of each individual computers clock can be off by up to 300 ms.   It just means that your local clock is ahead of the other peer by 23 ms plus the network latency in fetching the block.

So there is a margin of error of 300ms in latency numbers?

That makes it pretty tough to get an accurate picture. I thought ntp would take care of that, not so?

The built in NTP client pings an NTP server and updates an internal offset if the delta time between the request and response is less than 300 ms.   So if you can ping the NTP server in 0 ms then you can have up to 300 ms in error.   If your ping time to the NTP server is 100 ms, then you will only have 200 ms of error.

Technically it is possible to get a more accurate time sync than this.


Quote
NTP v4 with kernel mods to support it, is capable of much better than 1ms accuracy, possibly as good as 1ns. According to [Dave Mills] article, NTP v3 is accurate to 1-2ms in a LAN and 10s of ms in WAN nets. http://www.cis.udel.edu/~mills/ntp.html

Other articles suggest that with an accurate time source, such as a GPS time source, NTP is accurate to 50us, but the links on the Linux kernel support say that accuracy of a few ms are possible. http://www.atomic-clock.galleon.eu.com/support/ntp-time-server-accuracy.html

Another article says that it is dependent on the predictability of network delays (i.e. a low jitter network). http://www.postel.org/pipermail/end2end-interest/2003-April/002925.html

In other words, highly paid witnesses could configure their local clocks with higher quality NTP solutions and get down to under 100 ms.    It would be a lot of work to get the internal NTP to be more accurate.   
For the latest updates checkout my blog: http://bytemaster.bitshares.org
Anything said on these forums does not constitute an intent to create a legal obligation or contract between myself and anyone else.   These are merely my opinions and I reserve the right to change them at any time.

Offline Thom

Re: October 5 Test Network
« Reply #307 on: October 07, 2015, 07:25:04 pm »
I see. Thanks for the thorough explanation.

The accuracy of each individual computers clock can be off by up to 300 ms.   It just means that your local clock is ahead of the other peer by 23 ms plus the network latency in fetching the block.

So there is a margin of error of 300ms in latency numbers?

That makes it pretty tough to get an accurate picture. I thought ntp would take care of that, not so?

The built in NTP client pings an NTP server and updates an internal offset if the delta time between the request and response is less than 300 ms.   So if you can ping the NTP server in 0 ms then you can have up to 300 ms in error.   If your ping time to the NTP server is 100 ms, then you will only have 200 ms of error.

Technically it is possible to get a more accurate time sync than this.


Quote
NTP v4 with kernel mods to support it, is capable of much better than 1ms accuracy, possibly as good as 1ns. According to [Dave Mills] article, NTP v3 is accurate to 1-2ms in a LAN and 10s of ms in WAN nets. http://www.cis.udel.edu/~mills/ntp.html

Other articles suggest that with an accurate time source, such as a GPS time source, NTP is accurate to 50us, but the links on the Linux kernel support say that accuracy of a few ms are possible. http://www.atomic-clock.galleon.eu.com/support/ntp-time-server-accuracy.html

Another article says that it is dependent on the predictability of network delays (i.e. a low jitter network). http://www.postel.org/pipermail/end2end-interest/2003-April/002925.html

In other words, highly paid witnesses could configure their local clocks with higher quality NTP solutions and get down to under 100 ms.    It would be a lot of work to get the internal NTP to be more accurate.
Injustice anywhere is a threat to justice everywhere - MLK |  Verbaltech2 Witness Reports: https://bitsharestalk.org/index.php/topic,23902.0.html

Offline Thom

Re: October 5 Test Network
« Reply #308 on: October 07, 2015, 07:28:13 pm »
Would PTP be a viable option?
« Last Edit: October 07, 2015, 07:30:30 pm by Thom »
Injustice anywhere is a threat to justice everywhere - MLK |  Verbaltech2 Witness Reports: https://bitsharestalk.org/index.php/topic,23902.0.html

Offline bytemaster

Re: October 5 Test Network
« Reply #309 on: October 07, 2015, 07:29:22 pm »
bitshares-argentina  can you please let me know if you have changed your software in any way because it appears your node is consistently producing blocks even though you are not in the set of active witnesses.

Everyone else, please update your witness to the latest master before block 58500  (about 4 hours).

Hi dan, I didn't change anything. Neither played with uncommon settings as far as I remember . 

Running from master branch with last commit db84a492b9a2cf290f92a3987b59c11808138d56

How can I help debugging?

Can you post your config file and command line arguments. 
For the latest updates checkout my blog: http://bytemaster.bitshares.org
Anything said on these forums does not constitute an intent to create a legal obligation or contract between myself and anyone else.   These are merely my opinions and I reserve the right to change them at any time.

Offline rnglab

  • Full Member
  • ***
  • Posts: 171
    • View Profile
  • BitShares: rnglab
Re: October 5 Test Network
« Reply #310 on: October 07, 2015, 07:52:03 pm »
bitshares-argentina  can you please let me know if you have changed your software in any way because it appears your node is consistently producing blocks even though you are not in the set of active witnesses.

Everyone else, please update your witness to the latest master before block 58500  (about 4 hours).

Hi dan, I didn't change anything. Neither played with uncommon settings as far as I remember . 

Running from master branch with last commit db84a492b9a2cf290f92a3987b59c11808138d56

How can I help debugging?

Can you post your config file and command line arguments.

.
Code: [Select]
/witness_node -d oct5-2 --genesis-json oct5-genesis.json -s "104.236.51.238:2005"

Code: [Select]
# Enable block production, even if the chain is stale.
enable-stale-production = false

# Percent of witnesses (0-99) that must be participating in order to produce blocks
required-participation = false

# ID of witness controlled by this node (e.g. "1.6.5", quotes are required, may specify multiple times)
witness-id = "1.6.37"

# Tuple of [PublicKey, WIF private key] (may specify multiple times)
private-key = ["GPH7vyDvX1dxcPvXs63A4cctf8ZFRcb7a3QGWoQYdX6ckuKJonbyx","5Kxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx"]

updated but not producing blocks. Now when I dump_private_keys there's no priv key for the signing key. I'll try on a new wallet.
« Last Edit: October 07, 2015, 07:58:07 pm by rnglab »

Offline jtme

Re: October 5 Test Network
« Reply #311 on: October 07, 2015, 07:54:03 pm »
BM, I think there are now 3 witnesses waiting to be voted in:

bitshares-argentina
triox-delegate
pmc

Voted.   Any others?

yep, I think I got voted out by this (had only my votes) 1.6.31

Offline Bhuz

  • Committee member
  • Sr. Member
  • *
  • Posts: 467
    • View Profile
  • BitShares: bhuz
Re: October 5 Test Network
« Reply #312 on: October 07, 2015, 07:58:18 pm »
It isn't just a matter of getting high end servers, we need to actually change the P2P protocol before we can hope to achieve those speeds.

When you are not busy, could you read my post about witnesses signing blocks in General Discussion?

Offline clayop

  • Hero Member
  • *****
  • Posts: 2033
    • View Profile
    • Bitshares Korea
  • BitShares: clayop
Re: October 5 Test Network
« Reply #313 on: October 07, 2015, 08:00:36 pm »
60 TPS / 20 min test will begin at 21:00 UTC

Add: CORE donation to 'clayop' is more than welcome :D
« Last Edit: October 07, 2015, 08:33:49 pm by clayop »
Bitshares Korea - http://www.bitshares.kr
Vote for me and see Korean Bitshares community grows
delegate-clayop

Offline Thom

Re: October 5 Test Network
« Reply #314 on: October 07, 2015, 08:56:58 pm »
60 TPS / 20 min test will begin at 21:00 UTC

Add: CORE donation to 'clayop' is more than welcome :D

Sent you 4000 CORE
Injustice anywhere is a threat to justice everywhere - MLK |  Verbaltech2 Witness Reports: https://bitsharestalk.org/index.php/topic,23902.0.html