What Is Your Missed Ticket %

Discussion in 'Tickets' started by drunkenmugsy, Mar 10, 2016.

  1. 2017/12/15 - Decred v1.1.2 released! → Release Notes  → Downloads
?

How many revoked PoS votes have you had?

  1. 1

    8 vote(s)
    28.6%
  2. 2

    3 vote(s)
    10.7%
  3. 3

    6 vote(s)
    21.4%
  4. 4

    0 vote(s)
    0.0%
  5. 5

    0 vote(s)
    0.0%
  6. 6

    1 vote(s)
    3.6%
  7. 7

    2 vote(s)
    7.1%
  8. 8

    0 vote(s)
    0.0%
  9. 9

    1 vote(s)
    3.6%
  10. 10+

    7 vote(s)
    25.0%
  1. drunkenmugsy

    drunkenmugsy Sr. Member
    Advocate (Reddit)

    Dec 28, 2015
    405
    218
    Male
    #1 drunkenmugsy, Mar 10, 2016
    Last edited: Mar 10, 2016
    OK I am getting a bit peeved by my missed ticket percentage. Almost 10%! I am running on a RPi2 dedicated only to SOLO PoS. Running on an upgraded tier cable modem connection in the USA. I am trying to figure out if I need to move my PoS host to another connection or use another platform altogether. If you SOLO PoS please run a getstakeinfo with date/time and post up. Also post up ping stats for 10+ pings going to decred.org please. Yes I realize decred.org may or may not be where you are sending vote data to but it is a common address we can all use without getting to technical.
    It seems to happen only when a new block is found very quickly after I am called to vote. A less than 30sec block time after a vote seems to trigger a revoke almost every time.

    RPi2 CPU load 1/5/15min sits around .10 or less

    {
    "poolsize": 41768,
    "difficulty": 8.49962772,
    "allmempooltix": 6,
    "ownmempooltix": 0,
    "immature": 6,
    "live": 241,
    "proportionlive": 0.005769967439187895,
    "voted": 151,
    "totalsubsidy": 280.91069628,
    "missed": 12,
    "proportionmissed": 0.0002872187649593107,
    "revoked": 12
    }
    Thu 10 Mar 17:49:53 *USA* 2016

    Ping stats to decred.org
    --- decred.org ping statistics ---
    20 packets transmitted, 20 received, 0% packet loss, time 19026ms
    rtt min/avg/max/mdev = 38.806/53.674/140.183/27.130 ms


    I am curious what other RPi2 in particular users are seeing and what your setup including connection is. Other PoS miners please post up as well for comparison. What are you guys seeing using AWS?
     
  2. ceejep

    ceejep Sr. Member
    Developer

    Dec 14, 2015
    192
    220
    I would hazard a guess that the RPi daemon is just too slow. Running the wallet on the RPi is fine, but right now a block can take a second or two to validate on an x64 core. The code is very slow and unoptimized. You might want to run a dedicated daemon on a PC or a VPS.
     
  3. drunkenmugsy

    drunkenmugsy Sr. Member
    Advocate (Reddit)

    Dec 28, 2015
    405
    218
    Male
    That is what I am trying to get a feel for by asking others to post up. I can do that I just want to make sure it is a RPi2 issue and not a block being found to fast for anyone to vote fast enough issue.
     
  4. sambiohazard

    sambiohazard Sr. Member

    Jan 21, 2016
    844
    372
    I have 6/60 revoked but only 1 of them was due to a power outage, rest are from time when pools were not working properly with votes. Also i am on a PC.
     
  5. Ubik

    Ubik New Member
    Translator (Italiano)

    Feb 12, 2016
    49
    13
    Do missed tickets get deleted or do they get more chances to get selected?
     
  6. sambiohazard

    sambiohazard Sr. Member

    Jan 21, 2016
    844
    372
    Missed tickets are revoked/deleted and your funds - tx fee are returned to your wallet.
     
    Ubik likes this.
  7. drunkenmugsy

    drunkenmugsy Sr. Member
    Advocate (Reddit)

    Dec 28, 2015
    405
    218
    Male
    This is for solo PoS only. Pools would not depend on my RPi2 to vote.
     
  8. ClokworkGremlin

    ClokworkGremlin Sr. Member

    Jan 10, 2016
    535
    381
    Male
    Whatever I want.
    #8 ClokworkGremlin, Mar 10, 2016
    Last edited: Mar 10, 2016
    I have 3 missed and 83 voted. Speaking of which, the "proportionmissed" value is reporting as 0.00007 which is a bit lower than the 3.4% of total votes that I've actually missed.
    I'm also running on a RPi2B+ and I have another altcoin wallet, VNC service, and Mumble server running on the same machine. I could just be lucky.

    Maybe check ISPs as well? I have Century Link.

    I actually had a friend switch from Comcast to Century Link. Her download rate doubled, but her upload rate went down... but I tried to spectate her playing Sunless Sea both before and after the switch. It didn't work at all on Comcast, but on Century Link it was extremely performant, though the video quality was (unsurprisingly) low.
     
  9. root

    root Member

    Feb 3, 2016
    381
    76
    I had sth like "work not delivered fast enough" when mining on rk3188 on testnet. Stopped ARM experiments until ARM64 linux, but the time is my enemy, so many exciting things here to toy with ;)
     
  10. drunkenmugsy

    drunkenmugsy Sr. Member
    Advocate (Reddit)

    Dec 28, 2015
    405
    218
    Male
    On a side note my spendable shows 23dcr right now. Stake diff is 15.6 yet I am not buying another ticket. I even tried manually. It tells me "-4: not enough to purchase sstx". What gives? I turned on --prunetickets when I start my wallet. It seems that get spend is not accurate?
     
  11. jolan

    jolan Sr. Member
    Developer

    Dec 7, 2015
    197
    226
    Male
    Decred Team Member
    Totally unscientific test (data below). Clocks synced with NTP, no load other than dcrd, but different peer lists.

    The RPi2 definitely has the most lag of the bunch.

    As @ceejep said, the code isn't optimized so faster hardware is undoubtedly better. Another thing that needs to be taken into account is the connection path that the vote travels through (you<--->your peer list<--->solo miner/mining pool).

    So a fast path with an RPi2 is probably ok, but a slow path with an RPi2 is going to cause missed votes sometimes.

    One thing that might help is to check the power saving mode:

    Code:
    cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
    If it's not set to "performance" then do so:

    Code:
    sudo sh -c 'echo performance > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor'
    And probably put that in rc.local or whatever so it's permanent across reboots.

    That way there isn't any lag waiting for the kernel to realize there's work to do (process a block) and kick the cores out of power saving mode.

    i7-4790k:

    10:45:03 2016-03-11 [INF] BMGR: Processed 1 block in the last 12m35.4s (4 transactions, height 9498, 2016-03-11 10:44:16 -0600 CST)
    10:47:01 2016-03-11 [INF] BMGR: Processed 1 block in the last 1m58.04s (2 transactions, height 9499, 2016-03-11 10:46:07 -0600 CST)
    10:51:11 2016-03-11 [INF] BMGR: Processed 1 block in the last 4m10.05s (1 transaction, height 9500, 2016-03-11 10:50:58 -0600 CST)
    11:03:22 2016-03-11 [INF] BMGR: Processed 1 block in the last 12m10.96s (2 transactions, height 9501, 2016-03-11 11:02:26 -0600 CST)
    11:06:01 2016-03-11 [INF] BMGR: Processed 1 block in the last 2m38.98s (1 transaction, height 9502, 2016-03-11 11:05:56 -0600 CST)
    11:07:22 2016-03-11 [INF] BMGR: Processed 1 block in the last 1m20.5s (2 transactions, height 9503, 2016-03-11 11:07:09 -0600 CST)
    11:08:41 2016-03-11 [INF] BMGR: Processed 1 block in the last 1m19s (1 transaction, height 9504, 2016-03-11 11:08:38 -0600 CST)

    ODROID-C2:

    10:45:04 2016-03-11 [INF] BMGR: Processed 1 block in the last 12m35.5s (4 transactions, height 9498, 2016-03-11 10:44:16 -0600 CST)
    10:47:02 2016-03-11 [INF] BMGR: Processed 1 block in the last 1m58.51s (2 transactions, height 9499, 2016-03-11 10:46:07 -0600 CST)
    10:51:12 2016-03-11 [INF] BMGR: Processed 1 block in the last 4m9.57s (1 transaction, height 9500, 2016-03-11 10:50:58 -0600 CST)
    11:03:23 2016-03-11 [INF] BMGR: Processed 1 block in the last 12m11.28s (2 transactions, height 9501, 2016-03-11 11:02:26 -0600 CST)
    11:06:02 2016-03-11 [INF] BMGR: Processed 1 block in the last 2m39.06s (1 transaction, height 9502, 2016-03-11 11:05:56 -0600 CST)
    11:07:23 2016-03-11 [INF] BMGR: Processed 1 block in the last 1m20.14s (2 transactions, height 9503, 2016-03-11 11:07:09 -0600 CST)
    11:08:42 2016-03-11 [INF] BMGR: Processed 1 block in the last 1m19.35s (1 transaction, height 9504, 2016-03-11 11:08:38 -0600 CST)

    RPi3:

    10:45:04 2016-03-11 [INF] BMGR: Processed 1 block in the last 12m35.96s (4 transactions, height 9498, 2016-03-11 10:44:16 -0600 CST)
    10:47:02 2016-03-11 [INF] BMGR: Processed 1 block in the last 1m57.92s (2 transactions, height 9499, 2016-03-11 10:46:07 -0600 CST)
    10:51:13 2016-03-11 [INF] BMGR: Processed 1 block in the last 4m10.37s (1 transaction, height 9500, 2016-03-11 10:50:58 -0600 CST)
    11:03:23 2016-03-11 [INF] BMGR: Processed 1 block in the last 12m10.7s (2 transactions, height 9501, 2016-03-11 11:02:26 -0600 CST)
    11:06:02 2016-03-11 [INF] BMGR: Processed 1 block in the last 2m38.41s (1 transaction, height 9502, 2016-03-11 11:05:56 -0600 CST)
    11:07:23 2016-03-11 [INF] BMGR: Processed 1 block in the last 1m20.85s (2 transactions, height 9503, 2016-03-11 11:07:09 -0600 CST)
    11:08:42 2016-03-11 [INF] BMGR: Processed 1 block in the last 1m19.37s (1 transaction, height 9504, 2016-03-11 11:08:38 -0600 CST)

    RPi2:

    10:45:05 2016-03-11 [INF] BMGR: Processed 1 block in the last 12m35.04s (4 transactions, height 9498, 2016-03-11 10:44:16 -0600 CST)
    10:47:03 2016-03-11 [INF] BMGR: Processed 1 block in the last 1m58.25s (2 transactions, height 9499, 2016-03-11 10:46:07 -0600 CST)
    10:51:13 2016-03-11 [INF] BMGR: Processed 1 block in the last 4m10.16s (1 transaction, height 9500, 2016-03-11 10:50:58 -0600 CST)
    11:03:24 2016-03-11 [INF] BMGR: Processed 1 block in the last 12m11s (2 transactions, height 9501, 2016-03-11 11:02:26 -0600 CST)
    11:06:03 2016-03-11 [INF] BMGR: Processed 1 block in the last 2m38.69s (1 transaction, height 9502, 2016-03-11 11:05:56 -0600 CST)
    11:07:24 2016-03-11 [INF] BMGR: Processed 1 block in the last 1m20.56s (2 transactions, height 9503, 2016-03-11 11:07:09 -0600 CST)
    11:08:43 2016-03-11 [INF] BMGR: Processed 1 block in the last 1m19.44s (1 transaction, height 9504, 2016-03-11 11:08:38 -0600 CST)
     
    ec8_civic and drunkenmugsy like this.
  12. dave

    dave New Member

    Jan 11, 2016
    27
    10
    {
    "poolsize": 41692,
    "difficulty": 15.66621457,
    "allmempooltix": 0,
    "ownmempooltix": 0,
    "immature": 1,
    "live": 113,
    "proportionlive": 0.002710352105919601,
    "voted": 85,
    "totalsubsidy": 158.13504231,
    "missed": 1,
    "proportionmissed": 2.3984841580121363e-05,
    "revoked": 1
    }

    azure:
    Packets: Sent = 20, Received = 20, Lost = 0 (0% loss),
    Approximate round trip times in milli-seconds:
    Minimum = 19ms, Maximum = 22ms, Average = 19ms

    aws:
    Packets: Sent = 20, Received = 20, Lost = 0 (0% loss),
    Approximate round trip times in milli-seconds:
    Minimum = 21ms, Maximum = 22ms, Average = 21ms
     
  13. ClokworkGremlin

    ClokworkGremlin Sr. Member

    Jan 10, 2016
    535
    381
    Male
    Whatever I want.
    Mine is set to "ondemand," but I'll keep this in mind if I start missing an unusually large number of tickets.

    I believe it's worth noting that even with the chain server, wallet, control panel, additional altcoin wallet, and Mumble server running, monitoring through VNC, the CPU load hovers right around 4%. Memory load is closer to 80% and is almost entirely the QT wallet for the other altcoin.

    This is one of the reasons I keep saying the Pi3 should have had 2 or 4 GB of memory.
     
  14. drunkenmugsy

    drunkenmugsy Sr. Member
    Advocate (Reddit)

    Dec 28, 2015
    405
    218
    Male
    Interesting. Changed:
    Code:
    mugsy@DCRDNode:~ $ cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
    ondemand
    
    I dont think this is a device issue but more a network issue as one of the devs alluded to early in the PoS discussions. Maybe we need to look in to preferred nodes based on rtt of net traffic?
     
  15. David

    David Sr. Member

    Jan 22, 2016
    364
    207
    Male
    USA
    I've also noticed this when manually purchasing tickets... I have some speculations but I will hold them until I know for sure.
     
  16. drunkenmugsy

    drunkenmugsy Sr. Member
    Advocate (Reddit)

    Dec 28, 2015
    405
    218
    Male
    So there is a 2-3 second difference in the first time listed. If that is really making the difference then every PoS miner needs to be on a backbone link to the internet - not feasible.
     
  17. ceejep

    ceejep Sr. Member
    Developer

    Dec 14, 2015
    192
    220
    There was a pretty bad bug in dcrrpcclient that caused stake outputs not to be tracked on wallet restart.

    https://github.com/decred/dcrrpcclient/commit/b3f48780a0d68e24ef6e915e930a1c1e58b69810

    This will be fixed on the next build, in the meantime you can build on your own and try resyncing the wallet to fix.
     
    David likes this.
  18. Myagui

    Myagui Jr. Member
    Pool Operator (PoS)

    Jan 5, 2016
    46
    45
    Male
    Been wanting to post my figures here and respond to the questionnaire, but I'm not sure what to make of this difference between missed and revoked:

    {
    "poolsize": 42041,
    "difficulty": 8.34382528,
    "allmempooltix": 10,
    "ownmempooltix": 0,
    "immature": 4,
    "live": 152,
    "proportionlive": 0.0036155181846292904,
    "voted": 118,
    "totalsubsidy": 219.56919972,
    "missed": 2,
    "proportionmissed": 4.757034464714697e-05,
    "revoked": 1
    }

    Missed and revoked have been sitting on these numbers for a very long time now, well over a week. The only vote misses that I've had, were caused by power and/or internet outage around the time of my tickets getting picked. Crappy providers :(

    So: can anyone explain to me why I have 2 missed but only 1 revoked? I though that all missed tickets would get automagically revoked?
    Thanks!
     
  19. ceejep

    ceejep Sr. Member
    Developer

    Dec 14, 2015
    192
    220
    They may not if you're offline to receive the notification. You can force a retry with

    Code:
    dcrctl rebroadcastmissed
    It should call this on start-up, it's TODO
     
    adam2312 and brencelj like this.
  20. Myagui

    Myagui Jr. Member
    Pool Operator (PoS)

    Jan 5, 2016
    46
    45
    Male
    @ceejep : thank you, that sorted it :cool:
     

Share This Page