AreBelongToU$ ⚡

Someone set us up the bank #bitcoin #btc #hodl #blockchain #crypto #cryptocurrencies #ethereum #eth #altcoins #ico #cboe #cme #lightning #mining #litecoin #ltc

[AreBelongToU$ ⚡ @AllYourBanks]

@TraceMayer You also came out !! But not in a good way....

[AreBelongToU$ ⚡ @AllYourBanks]

@Crypt0Events @advbitcoin Ding ding ! 😃

[AreBelongToU$ ⚡ @AllYourBanks]

@phraudsta 2. Ah this is model based on SF Vs actual. Thanks ...

[AreBelongToU$ ⚡ @AllYourBanks]

@phraudsta 1. Ah your comment at this line could be interpreted at 'f' being daily flow, perhaps change the comment . 2. Price[price.length-1] != Exp(beta0)*Pow(sf[sf.length-1], beta1) 3. I keep getting 'cannot read property beta' when using jstat

[AreBelongToU$ ⚡ @AllYourBanks]

@phraudsta Wrt to the script : 1. Why is the flow multiplied by 365.25 ? 2. When I try to recompute the actual usd btc price from the ln(usd) price it doesn't match with the latest price from the array, is that expected ? 3. jStat is deprecated, jstat (low caps) replacment doesnt work..

[AreBelongToU$ ⚡ @AllYourBanks]

To Node or Node(al), thats the question !

[AreBelongToU$ ⚡ @AllYourBanks]

@ThinkingUSD Great minds ....

[AreBelongToU$ ⚡ @AllYourBanks]

@rootzoll @rubberroad @Xentagz @stephanlivera Read that text but it doesn't explain how consistency is achieved. Would have to look at the code but wont (sorry). Perhaps someone can trigger/check w LND team.

[AreBelongToU$ ⚡ @AllYourBanks]

@rootzoll @rubberroad @Xentagz @stephanlivera Better yet .. a quote :

linux/ext3: If the hd has write caching enabled, then the data may not really be on perm storage when fsync() returns

Linux/ext4: The fsync() impl in older kernels and lesser used fs's does not know how to flush disk caches. – issue adressed recently

[AreBelongToU$ ⚡ @AllYourBanks]

@rootzoll @rubberroad @Xentagz @stephanlivera Another nice article : https://t.co/CQTx2AsSxs

Or search for "fsync write barrier" if you dont trust my links ...

[AreBelongToU$ ⚡ @AllYourBanks]

@rootzoll @rubberroad @Xentagz @stephanlivera Conceptual : even when using database transactions, I assume that the "underlying" must give strong guarantees as well on preventing partial writes as much as possible else the promise on db level might not hold

[AreBelongToU$ ⚡ @AllYourBanks]

@rootzoll @rubberroad @Xentagz @stephanlivera Yes it really depends on the LND impl. It can for example affect the behaviour of fsync() afaik : with caching fsync() will return prematurely !

[AreBelongToU$ ⚡ @AllYourBanks]

@pierre_rochard @PatrickLemke @BR_Robin Tip btw : to minimize FS corruption i.c.o pow outtage, look at ways to disable "write caching" (google enable write barrier support). This degrades perf but gives less corr risk. For example you can use barrier=1 option for ext4 in a linux fstab (hd+kern needs to supp this too!)

[AreBelongToU$ ⚡ @AllYourBanks]

@pierre_rochard @PatrickLemke @BR_Robin Careful .. Raid 1 or alike is not a measure allowing you to skip backups ! It is only about redundancy mitigating HD failure. In general a 'traditional' backup is still needed to mitigate loss due for example filesystem corruption.

[AreBelongToU$ ⚡ @AllYourBanks]

@rubberroad @rootzoll @Xentagz @stephanlivera Typo, must be https://t.co/VPwrRR3BAP

[AreBelongToU$ ⚡ @AllYourBanks]

@rubberroad @rootzoll @Xentagz @stephanlivera Look at the faq at https://t.co/PvzqzS6zTy (search write cache). On raspberry I use ext4 with option barrier=1 option in fstab and my HD supports it. Performance will degrade but there less risk of corruption

[AreBelongToU$ ⚡ @AllYourBanks]

@rubberroad @rootzoll @Xentagz @stephanlivera Google "barriers and journaling filesystems" for a great LWN article about (journaling) filesystem corruption

[AreBelongToU$ ⚡ @AllYourBanks]

@rubberroad @rootzoll @Xentagz @stephanlivera Also: seems like btrfs is not well suited for lots of random writes (e.g databases). Again look at the faq or search for "write amplification risk" as it is a CoW type system. Disclaimer : I have no personal experience but when looking at it this is what struck me...