Every single node on IOTA’s network went down on July the 6th according to Sarah Lewis, a researcher. She said:
“Currently the confirmed transaction ratio in iota is practically 0 because someone is stitching the side tangle to the main tangle and this apparently breaks regular nodes ability to select tips…
This seems to have taken every single iota peer offline… I’ve done spot checks on like a dozen nodes now and none of them are responding to even simple getNodeInfo requests.”
What exactly happened here is unclear. Someone, or some team, is apparently effectively spamming the network, but with a twist of sorts.
The attacker is connecting the main tangle to a side tangle through “stitching.” Now that’s not quite something for grandma to do as you need command line and knowhow, but we’re told grandma’s transaction wouldn’t mistakenly end up on the sidetangle.
“The sidetangle isn’t considered in the tip-selection without stitching transactions because the walk starts at a milestone,” Luca Moser, an iota dev working on the Trinity iota wallet says before further adding:
“A transaction using the default tip-selection algorithm will not end up on the sidetangle.” Which suggests the funds themselves are not at risk, with this apparently being a node DDoS method. Moser says:
“The sidetangle doesn’t matter. Unless by stitching it to the maintangle the walk will go onto it and therefore overload the IRI node which causes the abrupt decrease of transactions per second in the network as no new tips get selected. However, the IF is working on a fix for this problem.”
IRI nodes are iota’s reference nodes, with IF being the Iota Foundation. Edward Greve, Head Of Engineering at the Iota Foundation, says:
“The current sidechain and syncing issues are a new phenomenon for the Tangle, and we’re taking the opportunity to acquire data about how the Tangle is responding and performing.
We will keep you posted and share new information as we are able to. Please understand, these new phenomena are not always obvious, and investigation takes time.”
That suggests they’re not quite sure of what exactly the problem is or how to address it as it might not be clear what part of the code can be called the bug, at least for now.
The network is, however, currently running, but all sorts of strange shapes are appearing on the tangle. Presumably one can stitch a third, fourth, or an endless number of side tangles, which may be a problem.
“The sidetangle alone doesn’t have any effect on the network (besides taking up hdd/ssd space),” Moser says, calling it a bug “with the stitching causing the tip-selection to overload the IRI node.”
They are currently having a global snapshot “to prune the database and allow smaller nodes to keep participating in the Mainnet. Otherwise, they may run out of space and fall off the network,” they say.
That might temporarily address the problem, but whether it can have a long term solution does remain to be seen.