Dial error on windows 7 #4512
Assignees
Labels
Milestone
Comments
Closed
|
@djdv could you try disabling reuseport (set the environment variable |
Stebalien
added
kind/bug
This was referenced Dec 20, 2017
|
@gdeverlant your errors appear to be unrelated to this issue. Please file a new one (or open a thread on https://discuss.ipfs.io as this doesn't look like a bug). |
|
@Stebalien Setting IPFS_REUSEPORT to false doesn't seem to have any effect for me, I set it, start the daemon (built from current master), and try to connect to a peer, but encounter the same error as before. I just ran this with Go 1.10 but I was getting the same results on 1.9.2 a few days ago. Promt 1 Promt 2 |
|
Got it. Is there any way on windows to see if there is an existing connection on that port/IP? Any way to see if a connection is established (wireshark?). @whyrusleeping can you reproduce in your VM? |
|
I'm pretty confident that nothing on this machine should be using either of these ports, nothing is configured to and nothing is showing for it, switching to 4001 didn't resolve it for me either. |
|
I'm getting this on linux as well. It can be readily reproduced with the go-libp2p echo example. I'm also seeing the |
|
Can someone fix this problem ASAP because we can't build and integrate our product to IPFS. |
|
it's the first time I'm using ipfs. I didn't used any other versions yet! |
|
I just did tried your suggestion and here is the problems: It is better than before but still there are lots of error messages. |
|
Unfortunately, that's an unrelated webui bug. The webui's peer display isn't optimised for thousands of peers.
…On December 27, 2017 10:36:09 AM PST, gdeverlant ***@***.***> wrote:
I just did tried your suggestion and here is the problems:
It is better than before but still there are lots of error messages.
The peers are listed, the webui works well but when I want to see the
peers the UI hangs and chrome sucks 50 % of my cpu.
|
Please post them. Unfortunately, we have a lot of noisy error messages (that you can mostly ignore). |
Not the issue. I believe the issue is that non-reuseport dialing has been broken by libp2p/go-tcp-transport#9. We now try to set the source port when dialing, even when not dialing with reuseport, which obviously fails if we're already using that port (e.g., listening on it). Fix incoming. |
|
Fix coming in libp2p/go-tcp-transport#20 |
|
Could one of you test /ipfs/Qma9vnYd4LV7DJkcsx4on3ezzuHT5JQcU2J2GaS3jmeKCt? It should fix this. |
|
@Stebalien |
|
I just tried your file /ipfs/Qma9vnYd4LV7DJkcsx4on3ezzuHT5JQcU2J2GaS3jmeKCt It loads with only 2 errors: But the hell load of error happens when I try to go in the webui and check the peers. Chrome sucks the CPU I checked the G:\ipfs folder's properties and it seems to be set to read-only
Hellraiser of errors: |
Those errors are "normal" and should go away in the near future.
That's because the webui has some really unoptimized code rendering all connected peers on a globe. For now, the recommended way to check which peers you're connected to is to run
That would explain that list of errors but I'm pretty sure we don't set the repo to read-only (unfortunately, I don't use windows so it's a bit hard for me to check). Did it start out read-only (did it work and then stop working) or did it suddenly become read-only? |
|
@gdeverlant There may be something else preventing you from writing to the datastore, without information on how ipfs was invoked and the permissions for the G: volume, it's hard to say. I believe the dht cast errors are already known, I think they are being demoted to warnings not errors. Someone correct me if this is wrong, please. |
|
I've moved this new discussion to #4528 (I like to keep issues clean). |
|
@Stebalien If you don't use Windows just go get a Virtualbox image and start playing with. https://developer.microsoft.com/en-us/microsoft-edge/tools/vms/ |
|
@Stebalien |
|
@djdv well thats another issue entirely. looks like a potential badger bug |
|
@djdv do you still have the repo from this failure? If so, how big is it and would you be willing to share it with us? The badger team wants to look at the badger files to try and debug the panic you posted. |
|
@whyrusleeping I still have it but it in an active repo, which has since been modified via browsing other users hashes and adding things myself. I'm willing to give some form of remote access to it or upload a copy somewhere else (whichever is easier) but it's relatively large (178GB), as such I will contact you directly via email about this when I get a chance later. |
|
2nd panic with above build |
|
Fixed. |
Closed
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Version information:
0.4.14-dev
Type:
Bug
Severity:
Critical
Description:
No peers in swarm. When dialing, @djdv gets:
Caused by something in #4453.
The text was updated successfully, but these errors were encountered: