Topic: BitcoinQueue.com ⏳ ★ Real-time Queue Charts 📈 (Read 2387 times)

Sorry about letting you down for so long, the site is coming back to life soon!

I am running a dedicated full node for collection purposes now, so more reliability is to be expected.

Thanks for your interest, stay tuned...

https://blockchain.info/unconfirmed-transactions  

I think it looks good and see the many queues transaction for immediate confirmed

At the moment, if you need quick settlement you'd better triple the usual fee.

You should go at least >100 satoshi / byte if you want to make it a matter of hours.

Damn that doesn't look good especially for businesses

The queue has returned, and it's uglier than before! It's been hovering over 60k almost for the last 2 days

https://www.bitcoinqueue.com/details/



https://www.bitcoinqueue.com/details/

We currently see the highest peak on the graph since the site started several months ago!



https://www.bitcoinqueue.com/

Sorry that I missed your last post until now. Yes, I can confirm that BitcoinQueue.com is using blockchain.info transaction feed until now, partly because running a full Bitcoin node on my VPS would be costly, and also because among all centralized sources I could choose from, it seems the most sensible and reliable to pick.

It could change in the future, the idea would be to merge several sources into the incoming feed.

Again we can see unconfirmed transaction build-ups occurring on a regular basis. Not much has changed in the transaction fee landscape for the last three months, which is partly why no significant update was done until now.



I am now resuming development of BitcoinQueue.com to include a very useful feature in my opinion, which is individual unconfirmed transaction lookup and statistics. That means you can fill in the transaction ID, and see where your transaction sits in the queue, the probability that it will be included in the next blocks, and other useful statistics.

The gap with bitcoinfees.21 does not seem to be closing.
Every price bracket is different?
(i.e. 21-30 sat, 2650 tx pending Bitcoinqueue.com but only about 200 tx pending on Bitcoinfees.21)

Bitcoinqueue's total transactions do seem to tally with Blockchain.info
Therefore, Blockchain.info mempool does not tally with Bitcoinfees.21 mempool either.
I had suspected Blockchain.info did not count zero fee transactions, but as Bitcoinqueue (taking from Blockchain.info) does have "some" zero fee, that does not seem to be the case.

Can you explain this difference?
(do you consider Bitcoinqueue "accurate" for seeing "the mempool", will it be "more accurate" if you tap directly into the network?)
Can fees be accurately calculated if "the mempool" is a nonspecific thing, different to each site?
(Blockchain.info are not publicly predicting fees of course, but do show "total" unconfirmed. Seems quite meaningless without scope)

Sorry if I'm struggling to express my issue simply and clearly. 

[UPDATE]

10 donators already, thank you very much

Many new features will be added before next week, including the following:

• Block information with evolution of stats with each block (requested feature)
• More details and statistical analysis of the mempool/unconfirmed queue
• Live overview with totals and average wait time

Stay tuned!

[UPDATE]

Now live table data is coming up, get a sneak peak here

http://www.bitcoinqueue.com/live.html

First, we need to agree that 72h is the comparison metric. This was only recently changed on bitcoinqueue.com (it was initially 4h, then 8h), i.e. on Friday afternoon (UTC). Then during the week-end, we lost almost 24h of data because the tx feed had died.

The reason of the failure is the following: My tx feed is using a websocket client that seems to keep the entire stream in memory until the websocket is disconnected. That caused the process to use always more RAM, first filling the swap and then killed by the kernel to preserve the system. Unfortunately, although all exceptions were caught within the program, there was nothing at that moment to restart the process.

The collection process is now protected by a shell (literally) to restart the process automatically, and I will update the process itself to stop after a 100,000 tx or so to work around the memory "leak" (stream as buffer, not clearing up). It will be easier than modifying the websocket client libraries, which is an external package (not my work).

So, the process was restarted this morning around 9.30am (UTC). If you add 72 hours to this, it means that it will only be since Wednesday that we will be able to compare bitcoinqueue.com data with bitcoinfees.21 and it will then be interesting if the gap has closed by then as it should be the cast.

Eventually, the tx data will be fed from my own internal source, but I have to work on this first: either run a full Bitcoin Core instance to feed the unconfirmed transactions, or tap directly into the network, but I have to make sure that tx are decoded the same was as Bitcoin Core would

The donations and continued interest over the project gives me more enthusiasm to develop the project further, of course

PS: our data now seems to be coherent with blockchain.info, which is temporarily used as feed source.

Hey Matt, me again!

Cool, I am only motivated by actually wanting to see this work.
I don't want to jump in too soon, but I have another question.

Why are there only about 800 transactions under <10 sat on your chart, and about 3700 transactions in total.
On bitcoin.info there are about 4100 transactions in total,
and on bitcoinfees.21 there are about 3500 transactions >10 sat, and 43000 10 sat and under?

You're not annoying me at all, your feedback is very much appreciated, thanks!

Of course, I'm still in the beta / experimental phase, but as you said, I'm getting there.

A huge and nasty bug has just been busted in the detail view, i.e. the >100 fee class wasn't summed up correctly (it was giving the average for all classes above 100 instead of summing them into one "more than" class).

Now the simple view and detailed view are reconciled. Phew

I thought something was wrong. Keep working, you'll get there!
Looking better again now, with a quick glance. (it was flat lining, not now)

I'll clear off if i'm annoying you now? 

The collection process (tx feed) was killed by the kernel over last night

I will be adding monitoring soon, stats have started rebuilding around 9am UTC this Sunday.

The sampling method, used for intervals 8h and higher, has been changed for the detailed view, and will be used for the simple view soon.

Stay tuned!

Ah, /details/..  I was missing something!

Good job. this will surely be a good info source after halving.

http://www.bitcoinqueue.com/details/

Am I missing something (or are you)?

I That live? I don't see it, just the old version.