That’s really unfortunate. I think the first log should have included some more hints as to what went wrong when dialing the bootstrap peer.
I can try to recreate, if you like? Yesterday all of my trials seemed to end the same way. I just left the last one going and found out in the morning that it actually was alive.
BTW, my machine just froze for some minutes when uploading about 550 MB’s, but it went up OK.
EDIT: Tried again. Now it just works, so so sorry about that
More success compared to previous testnet. Now I managed to get faucet (also prev one), upload a single file successfull (and receive enough quotes without timeout).
What does not work like previous network is that if I upload many files e.g. 89 files the whole upload fails and returns not enough quotes were returned. Would it be an option to upload these 89 files in smaller batches to prevent a too larger buffer building up and failing? Even sequential one-by-one would work (slowest and probably more overhead messages for the network).
Making payment for 3746 Chunks that belong to 89 file/s. Error: Failed to send tokens due to Network Error Not enough store cost quotes returned from the network to ensure a valid fee is paid.
Hmm, seems like a normal amount of retries there.
Looking at this more, I suspect this might have gone up over some churn, and the payment targets changed, so not enough folk actually stored this data.
We still need retries in here, (and repayments), which would hopefully smooth this out.
Unfortunately I can’t really confirm/deny this suspicion on this testnet due to not having logs on this path. I’ll add that now and we can hopefully pin this down going forwards
Looks like you’re missing the -n
parameter. We can flag that with a better error message there, I think
edit: -n
not --name
sorry
edit edit: this should make that error much more helpful: setup color eyre by joshuef · Pull Request #708 · maidsafe/autonomi · GitHub
you should share your updated script for node checking
thats me up to 110 nodes over two different vps’s
Strange though as it was running for around 10 hours for just those 11 files - downloading and uploading the whole time … if only those ten tries for that chunk, it seems there is something else happening IMO.
I’m going to clear everything out and start again with some new files and see if I can run it through nethogs to get a clear picture of the up/down amounts.
It’s for many chunks. I just pulled that one out as an example. I could see 5 different fails in that one file eg.
Up/down amount isn’t such a concern I think (if the upload process were working well, you would not be seeing the retrying…). I’d suggest keeping an eye on your logs and looking for retried PUTs and gets. That’ll give a clearer picture of what’s going on and where an error might
Will do, just left for Miami sorry, it will be when I get home this evening.
This is what I also got in the last test.
Do we have any idea what causes it?
This and the mem issue are my 2 party poopers of the moment.
I think it’s probably not paying the correct holders enough.
Which can be because the holders when we GetStoreCost
have changed (and so new have not been paid). OR, price has gone up.
There’s no fallback there as yet. Just having a poke at that the now
I see this every time I try to upload a larger file, say around 600MB. Small files around 20MB uploads just fine.
Error: Failed to send tokens due to Network Error Not enough store cost quotes returned from the network to ensure a valid fee is paid.
The StoreCost is likely an issue with a fee fix to reduce network load yesterday. I’ve undone some tolerance we had to fee payment, and that may well be causing us some issues here. fix(node): accept fees _near_ our current store cost. by joshuef · Pull Request #710 · maidsafe/safe_network · GitHub should sort that.
Ah interesting its large file related. That gives us something to go on there, cheers!
My sample size is just a few trials though… but seems that way.
You actually don’t need that. If you launch multiple node processes, each log directory will be unique because the peer ID will be used.
Why not pay for each chunk instead of bulk-billing? Is it that much more of a network load?
That’s what we do here and what enables the reward payments easily.
I vaguely remember a conversation about this, cost of moving them.
How many did you move at what cost?